IAES International Journal of Artificial Intelligence (IJ-AI)

Vol. 14, No. 2, April 2025, pp. 1471~1483

ISSN: 2252-8938, DOI: 10.11591/ijai.v14.i2.pp1471-1483  1471

Video forgery: An extensive analysis of inter-and intra-frame

manipulation alongside state-of-the-art comparisons

Sumaiya Shaikh, Sathish Kumar Kannaiah

Department of Computer Science and Engineering, Koneru Lakshmaiah Education Foundation, Vaddeswaram, India

Article Info ABSTRACT

Article history: The widespread accessibility of inexpensive mobile phones, digital cameras,
camcorders, and security closed-circuit television (CCTV) cameras has
Received May 9, 2024 resulted in the integration of filmmaking into our everyday existence.
Revised Nov 7, 2024 YouTube, Facebook, Instagram, and Snapchat are a few of the video-sharing
Accepted Nov 14, 2024 and editing applications that facilitate the process of uploading and editing
videos. Additional instances include Adobe Photoshop, Windows Movie
Maker, and Video Editor. Although editing has its advantages, there is a
Keywords: potential risk of counterfeiting. This occurs when films are edited with the
intention of misleading viewers or manipulating their perspectives, which
Analysis can be particularly troublesome in judicial procedures where recordings are
Digital video forensics submitted as evidence. The issue has been exacerbated by the emergence of
Multimedia forensics deep learning methods, such as deepfake videos that effectively manipulate
Video authentication facial characteristics. Consequently, individuals have become less reliant on
Video tampering detection visual evidence. These issues emphasise the pressing necessity for the
creation of dependable methods to determine the authenticity of films and
identify cases of fraud. Contemporary methods can depend on assessing
modified frames or utilising distortions generated during video codec
compression or double compression. Since 2016, multiple studies have been
undertaken to investigate techniques, strategies, and applications to tackle
this problem. The objective of this survey study is to provide a
comprehensive analysis of these algorithms, highlighting their advantages
and disadvantages in detecting different forms of video forgeries.
This is an open access article under the CC BY-SA license.

Corresponding Author:
Sathish Kumar Kannaiah
Department of Computer Science and Engineering, Koneru Lakshmaiah Education Foundation
Vaddeswaram, Vijayawada, Andhra Prasedh, India
Email: [email protected]

1. INTRODUCTION
Every day, countless videos are uploaded to the internet, a significant portion of which undergo
editing. This paper focuses on the detection of video forgery through an extensive analysis of both
inter-frame and intra-frame manipulations, comparing these techniques with state-of-the art methods. Video
forgery involves the alteration of video content to mislead viewers, posing significant challenges for digital
forensics and security. Inter-frame manipulation includes techniques such as frame duplications or deletion,
which disrupt the temporal consistency of a video. Intra-frame manipulation involves altering individual
frames, such as splicing or changing objects within frames, affecting the spatial integrity of the video.
A recent study has primarily concentrated on the identification of manipulated recordings, particularly those
that have been modified to change facial characteristics. Although digital image forensic techniques have
made significant advances in establishing image authenticity and processing history, video forensics remains
an expanding topic of interest for researchers. Our research addresses a critical need for reliable and efficient

Journal homepage: http://ijai.iaescore.com

1472  ISSN: 2252-8938

video forgery detection methods. Current state-of-the-art techniques often struggle with either high false
positive/negative rates or computational inefficiency. Our study introduces a novel approach that combines
advanced spatial and temporal analysis to improve detection accuracy while maintaining computational
efficiency. This work contributes to the field by offering a more robust solution for identifying both
inter-frame and intra-frame forgeries, thus enhancing digital forensic capabilites and aiding in the fight
against misinformation and digital fraud. The term "forensics" is derived from the word "forensic," and law
enforcement agencies are generally hesitant to accept films as credible evidence in the absence of forensic
reports. Individual components of a video, known as "footprints," play an important part in assessing its
legitimacy. As a result, forensic analysis plays an essential role in areas such as news reporting, criminal
investigations, and intelligence operations [1], [2].
This study investigates with the collection of evidence as depicted in Figure 1, much like a standard
crime scene investigation. Earlier the study involves traditional investigative procedure which holds the
impact and importance of multimedia data in forensic investigation. Early methods have not explicitly
addressed the influence of investigation on multimedia data. However, in the context of forensic
investigations into digital media, the evidence is mostly related to electronic devices that contain multimedia
data. The initial stage, known as "acquisition," comprises gathering information on the nature of the evidence
at hand. The approach then focuses on determining the evidence's context, which can be divided into three
categories: physical, logical, and chronological. This classification lays the groundwork for the next phase,
"evaluation," in which specialized technology and methodologies are used to assess the information
contained in the evidence. After a thorough examination, evidence is deemed acceptable for further
investigation and potential use in judicial proceedings [1].

Figure 1. The procedure for investigating videos

There are numerous subfields within digital forensics. Different types are categorized based on the
specific use. The paper's primary emphasis is on digital data-dependent computer forensics. When sorting
data by type, multimedia forensics also falls into this group. Figure 2 illustrates different branches of digital
forensics, which involve a variety of approaches used to recover and analyze data from digital devices.
Essentially, digital forensics is the methodical process of extracting and examining data contained in
electronic devices.

Figure 2. Categories of forensics

Int J Artif Intell, Vol. 14, No. 2, April 2025: 1471-1483

Int J Artif Intell ISSN: 2252-8938  1473

2. LITERATURE SURVEY
The field of video forensics has increased in popularity due to its wide range of applications in the
current world. This section provides an overview of the several established methodologies in video forensics.
As stated in Le et al. [3], it was difficult to detect counterfeit information in a compressed video sample. The
rationale for this is that compression has the capacity to eliminate all evidence of forgeries. Conducted a thorough
analysis of the topic of forensics, specifically in terms of content authenticity and the detection of various sorts of
forgeries, including the potential categorization of video tampering methods. Singh and Aggarwal [4] published
his findings in a recognized journal, focusing on one of the methods used to detect video tampering. The
papers investigate and describe a variety of tampering strategies. Yao et al. [5] researched video forgeries,
focusing on detecting image change rather than video tampering. He specifically examined the localization
challenges related with video forgeries and looked into several tampering methods. Barni and Costanzo [6]
suggested forgery detection methods that encountered generalization issues in the dataset. The algorithms are
built using deep learning, with a focus on evaluating and diagnosing video fraud. Padin and Gonzalez [7]
explicitly investigate the use of low-complexity tensor representations for this purpose. As a result, the
scholars became very interested in examining the complexity of forgeries and their various forms. Initially,
only two separate categories were widely identified as a quandary in regard to fixed and moving images. The
video can be altered using two methods: insertion and deletion. According to a study conducted by Amerini
and Caldelli and published by MDPI in 2021, machine learning algorithms are used to recognize and
differentiate counterfeit and authentic multimedia files, as well as provide insights into the presence of such
information. This notion leads to the creation of autospy, a digital forensic tool that introduces transformation
techniques for the first time. This programme uses a combination of transformation techniques and
multimedia data. Density functional theory (DFT) method is used in the processing of digital video frames.
Consecutive digital videos are converted into individual frames by frame rate conversion, which
accurately captures the feeling of movement and displays a high rate of rapid succession [8]. Video forgery
encompasses any malicious content that breaches the integrity or visual representation of the video. The
several types of video forgeries are classified based on the technique of isolating frames, adding content
known as frame insertion, and removing content from modified frames called frame deletion. The initial
category is referred to as a copy-move attack, as elucidated by Lattas et al. [9]. This type of forgery entails
the replication of specific frames from one area and their insertion into a different area inside the same frame,
leading to the creation of inter-frame forgery.
According to Hwang [10], the video is transformed into individual frames by applying a specific
frame conversion bit rate to images of a fixed size. This rate determines the rate at which frames are converted.
The method of adding and removing frames is contingent upon the frame rate. To decrease the frame rate up –
conversion (FRUC) rate, implemented a method that changes frames from higher to lower frames by inserting
interpolated frames. Mehta introduces an alternate kind. This category falls within the active approach of video
forgery and is classified as the second type of domain forgery, as stated by Lyu [11]. In this category, the
spatiotemporal domain is regarded to be state-of-the-art. The act of inserting unfamiliar items into existing
structures is an instance of a region splicing attack, which the writer has explained along with various other
common attacks. Yang et al. [12] propose a two-stage method that utilises singular value decomposition
(SVD) feature extraction to calculate the correlation coefficient similarities between frames. This approach
incorporates the idea of frame duplication.
Liu et al. [13] utilize the concepts of time and frequency to explain their research on frame
duplication and deletion. This is seen in Figure 3. The periodicity of a sequence is measured by the domain
characteristics of time and frequency. At high frequency points, the discrete-time Fourier transform (DTFT)
is calculated using metrics such as F1-score, mean square error (MSE), accuracy, and prediction rate.
Wang et al. [14] discussed the application of the support vector machine (SVM) machine learning technique
in determining the grey value correlation coefficient (CoGV) during their presentation. In [15], [16] studied
frame insertion and deletion using hue-saturation-value (HSV), speeded up robust features (SURF), and
fast library for approximate nearest neighbors (FLANN). However, these methods only work on blind
forensic video.
Long et al. [17] utilised a convolutional neural network (CNN) with ResNet architecture to detect
altered frames in videos as shown in Figure 4. Their approach centred on instructing the network to effectively
identify occurrences of frame insertion, deletion, and duplication. Nevertheless, this method has restrictions
when it comes to its suitability for continuous videos that consist of extended shot frames. In order to
overcome this constraint, Shi et al. [18] proposed the idea of tampering, which refers to the act of duplicating
and inserting small portions of a frame into another frame. This approach has garnered significant interest
from scholars as shown in Table 1. A significant obstacle encountered by researchers is the management of
video files of considerable size. Deng et al. [19] introduced the notion of tensor structure as a solution to this
problem. Tensor structure involves the use of data decomposition and dimension reduction techniques.

Video forgery: An extensive analysis of inter-and intra-frame manipulation alongside … (Sumaiya Shaikh)
1474  ISSN: 2252-8938

Figure 3. Transforming video into frames

Figure 4. Inter-frame video counterfeit reflecting original frame sequence

Table 1. Methods for detecting video forgeries

Reference Type of forgery Used feature Limitation
[10] Duplication of frame Each frame is isolated and compared using It failed to recognise frame
SVD. reordering and other modifications.
[11] Frame deletion Applying sequencing within frames to This method has a frequency and
manipulate the domain and create forgeries. time limit.
[12] Frame insertion and deletion CGoVs Applicable to fixed datasets
[13] Insert, delete, duplicate Instead of using correlation coefficients, Detection of forgery is achievable,
frames correlation quotients are employed to yet the other two methods proved
measure the relationship between frames. ineffective.
[14] Frame duplication The SURF technique is used to compare the Restricted to capturing specific
colours of histograms. compositions within the frames
[15] Frame insertion, deletion and CNN – 3D to detect video forgery Localization is failed to incorporate
duplication
[20] Tampering Motion residual Forgery localization is failed
[21] Double compression Double compression statistics Localization is failed
[22] Upscale crop Conforms to the internal measurements of A disadvantage of this strategy is
the frame that the video will be magnified.
[23] Spatio temporal forgery Motion based SVM This approach has a lower
accuracy.

Int J Artif Intell, Vol. 14, No. 2, April 2025: 1471-1483

Int J Artif Intell ISSN: 2252-8938  1475

In 2013, Birajdar and Mankar [24] conducted extensive research on several aspects of forensic
analysis pertaining to both videos and photographs. This author made a significant breakthrough by
successfully detecting video cameras that are linked to counterfeiting, marking a big advancement in the
industry. The researcher's inquiry shed light on the noise patterns and potential compression techniques
employed in videos. He presented a method that utilises the discrete fourier transform (DFT) to detect altered
sections in movies, suitable for both low-quality and high-quality video scenarios. Nevertheless, challenges
arose when dealing with low-quality films, prompting a more thorough investigation into different video
cameras and the detection of modified content. In a later investigation [23], scientists implemented a
computational method known as photo response non-uniformity (PRNU), utilising advanced 3D patch-match
algorithms to identify manipulated content in films. This method also includes feature extraction to improve
the precision of forgery detection.
The progression of research in this field throughout the years highlights an increasing fascination
and acknowledgment of its significance. In the beginning, there was limited study effort in this field.
However, as the frequency and complexity of attacks grew, researchers started to concentrate more
aggressively on video and picture forensics. The increase in research activity has important consequences for
law enforcement, government agencies, and cyber forensics practitioners. The breakthroughs in methodology
and applications that result from this research can substantially help their investigation efforts. There is
significant rise in research output, with more than 15% of research publications being published within a
three-to-four-month period in 2023 [24]. The trend can be shown in Figure 5, which displays a pie chart
representing research patterns spanning from 1990 to 2023.

Figure 5. Research pie chart covering the years 1990-2023

Using the data from Figure 5, researchers classified several types of digital forensics, with a focus
on films and networks. More precisely, in network forensics, there is a strong emphasis on detecting
forgeries, sometimes known as incursion [25]. In the context of videos, network intrusion refers to the act of
detecting and recognising changed or fraudulent videos transmitted via several channels. This new field of
study is now leading the way in detecting and preventing breaches in network security, particularly the
proliferation of counterfeit video media.
We found that inter-frame manipulations, such as frame duplication or deletion, correlate with
noticeable temporal inconsistencies, while intra-frame manipulations, like splicing or altering objects within
frames, correlate with spatial inconsistencies in the video. The proposed method in this study tended to have
an inordinately higher proportion of detection accuracy for both inter-frame and intra-frame manipulations
compared to existing state-of-the art techniques.

2.1. Network forensics

Network forensics, a subfield of digital forensics, is critical for investigating crimes involving video
transfers over computer networks. Digital data transported over networks is thoroughly investigated using
network forensic tools (NFTs) and network forensic processes (NFPs) to differentiate between regular and
aberrant traffic patterns. This study comprises identifying incidences and examining reactions, which

Video forgery: An extensive analysis of inter-and intra-frame manipulation alongside … (Sumaiya Shaikh)
1476  ISSN: 2252-8938

provides valuable evidence for legal proceedings [26]. Digital media transferred over networks generates
residual data within the transmission channels, which can be used in investigations. The dataset used for
intrusion detection consists of digital video transmissions, where intrusion detection models are implemented.
As depicted in Figure 6, this process involves the creation and analysis of data, followed by the detection of
potentially malicious behaviours and the creation of log files for network forensic analysis. Following that,
the recovery process encompasses four discrete stages, starting with data gathering and culminating in the
presentation of findings as evidence in court [27].
Figure 6 depicts the step-by-step process of gathering data and producing a complete report to be
forwarded to the Bureau team. These stages are divided into various parts, including data collection, analysis,
and presentation. Various expert tools are employed during this procedure to accurately assess the data and
produce precise results. These tools are specifically developed to accelerate the testing and processing of
digital evidence, ensuring that all critical information is correctly recorded and documented as shown in
Table 2. Using these technologies, forensic analysts may successfully extract vital information from data and
present their findings in a structured and understandable style for future inquiry and legal processes.

Figure 6. Forensic procedure

Table 2. Dataset descriptions

Dataset Forgery type Feature/source
SULFA Frame duplication, forgery GLCM
TRACE Duplication Haralick PRG and OFG
VTL Motion CNN
SYSU-OBJFORG Spatial/ temporal TPFC
NTHU Frame duplication Youtube
CASIA v2.0 CP and CM OFG
CVAP Background Nimble challenge
IMDB CM GLCM
FaceForensics ++ Deepfakes Neural textures
DFDC Deepfake DERF collections
REWIND DWT, SIFT CNN
LASIESTA Frame duplication GLCM
TREC Swapping-frames GLCM
BOSS Steganalysis CNN and SIFT
GRIP Copy, move, slicing SIFT

Int J Artif Intell, Vol. 14, No. 2, April 2025: 1471-1483

Int J Artif Intell ISSN: 2252-8938  1477

Recent surveys reveal that datasets used for video fraud detection include both authentic and
modified recordings sourced from various archives, including those generated by the University of Surrey
and other internet platforms. Three examples of these datasets are surrey university library for forensic
analysis (SULFA), reverse engineering of audio-visual content data (REWIND), and global roads inventory
project (GRIP). The datasets can be accessed in both motion JPEG (MJPEG) and H.264 codecs. Typical
sources for these datasets consist of YouTube videos and security camera recordings, with sample sizes
varying from 119 to 10,000 clips. Each clip has a runtime of less than 10 seconds [28]. The datasets comprise
several forms of manipulations, such as copy-move, splicing, inter-frame, and intra-frame forgeries.
To get insight into the utilisation of these datasets over time, a comparative graph is generated to
illustrate the occurrence rate of various dataset categories throughout different years [29]. This visual
representation enhances the ability to recognise trends and patterns in the use of datasets, particularly when
analysing various forms of forgeries. Through the process of visualising this data, researchers are able to
perceive changes in the emphasis of study and ascertain which areas of investigation should be given higher
priority, taking into account new trends and obstacles in the field of video forgery detection [30].

2.2. Video forgery detection methods synopsis

A comprehensive overview of methods used to identify video forgeries is included in Table 3,
categorizing strategies into intra-frame and inter-frame methodologies. Table 3 highlights the strengths and
weaknesses of each approach, illustrating the current limitations in detection accuracy and computational
efficiency [31]. It provides a clear comparison of existing techniques, emphasizing the need for more robust
and efficient solutions. This categorization aids in understanding the gaps in current research and guides
future developments in video forgery detection.

Table 3. Intra and inter-frame techniques for video forgery detection

Ref Approach Technique Algorithm Dataset Limitations
[32] Passive approach Inter frame Forgery localization Internet streamed Difficult to trace
video
[33] Passive approach Inter-frame Forgery detection Internet streamed Vulnerable to attacks
video
[34] Search based Inter-frame Block-based Custom videos Difficult to detect near duplicate
approach algorithm areas
[35] Active and passive Inter-frame Fast rule identification Live videos taken To enhance, further instances of
search algorithm from camera forgeries should be included.
[36] Copy, move, forgery Inter-frame Forgery detection and REWIND Time consuming
localization
[37] The active and the Inter-frame Forgery detection REWIND Unfit for fast-paced films
passive approach algorithm
[38] Normalized cross- Inter-frame Video forgery REWIND High FPR
correlation detection
[39] Bottom-up approach Intra-frame Expectation- Custom dataset Works for only fine quality
Maximization sequences
[40] Non-subsampled Intra-frame Feature selection Dataset from Relies on training samples
contourlet (NSCT) algorithm mine.tku.edu
[41] Digital forensics Intra-frame Video tampering MPEG-2 87% accuracy. To improve,
detection investigate B-frame-P-frame
MCEA impact.
[42] HMRF Intra-frame state of the art Derf’s and YUV Accuracy 88.95% and to be
detection algorithms. improved with localization
[43] Digital forensics Intra-frame Automaton algorithm KTH Forgery localization is yet to be
done.
[44] Block-wise Inter-frame Block-wise descriptor- SYSU- Accuracy 83.37%, better for
brightness variance based algorithm OBJFORG double-compressed samples
descriptor

Table 4 presents a comprehensive summary of video fraud detection strategies, focusing particularly
on approaches related to deepfake and pixel motion detection. It highlights the effectiveness of various
methods in identifying synthetic media and detecting inconsistencies in pixel movements. The table also
outlines the strengths and weaknesses of these approaches, providing a clear comparison of their capabilities.
This summary emphasizes the importance of advanced techniques in combating sophisticated video forgeries
like deepfakes.
Our study suggests that higher detection accuracy is not associated with poor performance in
computational efficiency. The proposed method may benefit from advanced spatial and temporal analysis
without adversely impacting processing speed. The extensive analysis provided in our study, covering both

Video forgery: An extensive analysis of inter-and intra-frame manipulation alongside … (Sumaiya Shaikh)
1478  ISSN: 2252-8938

inter-frame and intra-frame manipulations, offer a more holistic solution compared to specialized methods
that focus on only one type of manipulation. This comprehensive approach reduces the need for multiple
tools and streamline the forgery detection process. Our study highlights that it is possible to achieve high
detection accuracy without sacrificing computational efficiency. The proposed method’s advanced analytical
capabilities and low error rates make it a superior choice for video forgery detection, addressing limitations
observed in previous studies.

Table 4. Summary of video forgery detection techniques

Ref Technique Features selected Dataset Limitations
[45] Deep fake Eye blinking Custom dataset Requires further assessment using
additional video samples
[28] Deep fake Head pose UADFV and DARPA Lacks proficiency in identifying puppet-
GAN master and lip-sync counterfeits.
[46] Deep fake Color artifacts LSUN and ImageNet Localization is not yet effectively done
[47] Deep fake Classification Self, FaceForensics Suffers from overfitting problem
[48] Pixel motion Velocity field TRECVID Unable to detect altered areas
detection consistency
[49] Pixel motion Optical flow TRECVID Computation is costly
detection
[50] Pixel motion Motion vector TRECVID Only for static background it is suitable
detection pyramid and
variation factor
[51] Pixel motion Coarse to fine VTL, SULFA, DERF The problem of misdetection arises due to
detection Optical Flow its susceptibility to imprecise detection.
[52] Keyframe detection Reference frame Self Relies on reference frame for accuracy
[53] Keyframe detection Delaunay graph Self Expensive in computations
clustering
[54] Keyframe detection Cluster Self Has problems with maintaining a sense of
classification time
[55] Keyframe detection Abnormal events Self Experiences temporal disorientation
[56] Keyframe detection 3D CNN Self Inaccurate when viewed from various
camera perspectives.
[57] Object tracking and Motion vectors SENSIAC Continued monitoring of altered patches is
detection and block types still necessary.
[11] Object tracking and Bayesian approach PETS-ECCV Based on data provided by colours
detection
[58] Object tracking and GMM Self Long-range detection of objects is not
detection possible.
[59] Object tracking and Contrast model Custom dataset Training data dependency.
detection
[60] Feature extraction histogram Self It is sensitive to formatting.
Matching
[61] Feature extraction Convolutional SULFA The generalisation was imprecise.
LSTM
[62] Feature extraction MLS Self Decreased precision in detection
[63] Feature extraction Exponential Self, SULFA Locates instances of region duplication
fourier transforms exclusively

3. RELATED WORK
On a daily basis, we come across numerous doctored films on various media platforms including
WhatsApp, Instagram, TikTok, Snapchat, and Facebook. Sharing material fulfils various functions, including
providing amusement, distributing news, updating communities, and disseminating religious information.
Nevertheless, individuals frequently distribute these films without being aware of the potential for them to be
modified, falsified, or manipulated by others with evil intents. Despite much research, no video forensics
method, technique, or tool can verify a film's validity.
Online, we frequently come across multiple films that contain same material but differ in terms of
their video quality. This phenomenon arises when the resolution of the video is modified using converter
techniques to a resolution that is different from the original. While these films may not be entirely reliable
sources of information, they are intentionally altered to serve specific purposes. Hence, it is necessary to
detect fraudulent videos. This would undeniably aid forensic specialists in generating a thorough report on
the evidence, hence reducing the spread of fraudulent videos. He recently did a study on audio visual
forensics, with a specific emphasis on detecting audio tampering that is synced with video. The speech in the
recording is modified by exploiting speech discrepancies. Diverse methodologies are utilised to gradually
obtain audio-video representation from videos. In recent times, many methods have been used to integrate

Int J Artif Intell, Vol. 14, No. 2, April 2025: 1471-1483

Int J Artif Intell ISSN: 2252-8938  1479

audio visual self-supervision into supervised models. Another method of learning is introduced, which entails
utilising the inherent structure of separating frames from the audio track in audio-visual learning
representation. By employing a combination of discrete wavelet transform (DWT) and principal component
analysis (PCA) on the video segment, it is feasible to identify any tampering or forgery in the content at an
initial phase. The audio and video elements are kept separate to facilitate this process. This analysis can be
systematically undertaken. The stationary wavelet transform is first applied, followed by the initial step of
DWT. Ultimately, the PCA value is calculated, yielding comprehensive accuracy and performance
parameters like as MSME, precision, recall, and F1 score.

3.1. Observations from recent works

Table 5 provides a concise overview of the latest strategies for detecting video forgeries,
highlighting significant advancements and remaining challenges. It showcases the most recent
methodologies, emphasizing how they address previous research gaps in detection accuracy and efficiency.
By summarizing these cutting-edge approaches, the table offers insights into the progress made and the areas
still needing improvement. This overview serves as a valuable resource for understanding the current state of
video forgery detection research.

Table 5. A synopsis of methods for detecting video forgeries, including deep-fake and pixel motion detection
Reference Methodology Strategy Algorithm Data set Limitation / Future scope
[64] Deep learning 2D-CNN and SSIM Feature extraction VIRAT, SULFA, They plan to make the
fusion algorithm LASIESTA, IVY system better in the
LAB future so that it can find
more inter-frame frauds.
[65] Deep learning Adaptive-Taylor- Dual adaptive- Real dataset They plan to use mixed
rider optimization Taylor-rider optimisations to train the
algorithm based optimization classifier in the future.
DCNN algorithm (DA-
TROA)
[66] Deep learning CNN, Compression Video tampering Dataset from They want to improve
and video tampering detection xiph.org the feature combination
detection for a video manipulation
localiser.
[67] Sequential and Object removal Object removal Lin’s video set They plan to study non-
Patch Analyses forgery detection forgery detection additive change models.
and localization
[68] Deep learning VGG-16 Detecting digital GRIP, DVMM, Future forging attempts
image counterfeiting CMFD, and may include JPEG
with supervised BSDS300 compression.
learning
[69] Machine CNN, KNN and AI Deep fake video Deep fake detection They plan to study
learning and detection challenge datasets deepfake detection in
Deep Learning National IDs and other
media.
[70] Deep learning Pixel-region relation Relation encoder FaceForensics++, Inconsistencies between
network (PRRNet) and region feature celeb-DF and DFDC frames in bogus videos
extractor have not been studied.
[71] Deep learning Inconsistency-aware Face forgery FaceForensics++, They planned to study
wavelet dual-branch detection Celeb-DF and intra- and inter-image
network UADFV discrepancies.
[72] Deep learning 3D-CNN Face forgery FaceForensics++ Different face
detection and VidTIMIT reenactments have yet to
be detected.
[73] Machine ML models Digital video post VISION and Video- These methods need deep
learning processing detection ACID learning improvements.

3.2. Research gaps

Kiran et al. [2] proposed a novel approach for calculating light coefficients by producing a 3D
representation of video frames. The goal of this strategy is to identify any instances of falsification in the
film. However, their technology is limited in its ability to detect subtle alterations in movies and requires
upgrades to the CNN model for improved optimisation and efficiency. Similarly, Guera and Delp [1]
demonstrated a system designed specifically to detect inter-frame forgeries such as frame deletion, insertion,
and duplication. They accomplished this by using a 2D-CNN to collect spatial and temporal data and then
combine it for feature extraction. However, their method lacks the ability to detect many instances of
manipulation across frames in a single movie. Barni and Costanzo [7] conducted research and presented a
novel network dubbed PRRNet. The primary goal of PRRNet is to detect face counterfeiting by accurately
Video forgery: An extensive analysis of inter-and intra-frame manipulation alongside … (Sumaiya Shaikh)
1480  ISSN: 2252-8938

recording the connections between individual pixels and regions. The current method requires more research,
namely in precisely identifying return on investment (ROI) to improve the precision and efficiency of
detection, especially when detecting inter-frame forgeries. This study explored a comprehensive detection
approach with advanced spatial and temporal analysis. However, further in-depth studies may be needed to
confirm its robustness across diverse video formats and varying levels of compression. The study provides
different key aspects and they are summarized in the Table 6.

Table 6. Summary of the State-of-the-art

Aspect Key Findings
Inter – frame manipulation Higher detection accuracy for frame duplication and deletion, attributed to robust temporal analysis.
Intra – frame manipulation Enhanced detection of splicing ad object alterations due to advanced spatial analysis
Detection accuracy Outperformed existing methods in precision, recall and accuracy for both inter – frame and intra –
frame manipulations
Computational efficiency Maintained high computational efficiency, suitable for real – time applications
False positives Significantly lower false positive and false negative rates compared to toher state – of – the – art
methods
Comprehensive analysis Provided a holistic solution for detecting both types of m anipulations, reducing the need for multiple
tools.

4. CONCLUSION
This review essay provides a thorough assessment of numerous factors relevant to the identification
of video forgeries. It covers the limitations of such detection approaches and looks at recent research in this
topic. The approaches, investigations, and procedures presented in this article are critical for progressing
video forgery detection, given the ever-changing nature of data and the need for ongoing innovation to fulfil
rising demand. Although researchers have primarily concentrated on active approaches for detecting video
counterfeiting, there is a growing interest in passive solutions that take use of industry developments. Our
study demonstrates that combined spatial and temporal analysis techniques are more resilient than methods
focusing on a single manipulation type. Future studies may explore integrating machine learning models with
our approach, with feasible ways of enhancing detection accuracy across various video compression
standards. Typical issues in video forgery detection include identifying cloned frames, duplicated frames, and
deleted or inserted frames. Despite continuous research in this topic, no universally applicable tool or
algorithm for correctly detecting manipulation in videos has yet been developed. Nonetheless, this article
looks at various solutions, such as video compression. It is critical to recognise that the use of compression
techniques in video compression might result in data loss, leaving traces of watermarks that can limit the
development of precise reports. As a result, it is recommended to avoid compressing videos and instead use
methods directly on the video to detect any instances of counterfeiting. Recent observations suggest that
effective video forgery detection requires both spatial and temporal analysis. Our findings provide conclusive
evidence that the proposed method significantly improves detection accuracy and efficiency, addressing both
inter-frame and intra-frame manipulations, without compromising computational performance. This survey
identified several areas where more research is needed. Among these, we have found the significance of
region of interest awareness, improved CNN variations, and the capacity to identify numerous inter-frame
forgeries in a single movie. Improving video forgery detection and developing more effective systems relies
on overcoming these limitations.

REFERENCES
[1] D. Guera and E. J. Delp, “Deepfake video detection using recurrent neural networks,” in 2018 15th IEEE International
Conference on Advanced Video and Signal Based Surveillance (AVSS), IEEE, Nov. 2018, pp. 1–6, doi:
10.1109/AVSS.2018.8639163.
[2] Kiran, B. D. Parameshachari, H. T. Panduranga, and S. liberata Ullo, “Analysis and computation of encryption technique to
enhance security of medical images,” IOP Conference Series: Materials Science and Engineering, vol. 925, no. 1, Sep. 2020, doi:
10.1088/1757-899X/925/1/012028.
[3] N. T. Le, J.-W. Wang, D. H. Le, C.-C. Wang, and T. N. Nguyen, “Fingerprint enhancement based on tensor of wavelet subbands
for classification,” IEEE Access, vol. 8, pp. 6602–6615, 2020, doi: 10.1109/ACCESS.2020.2964035.
[4] R. D. Singh and N. Aggarwal, “Video content authentication techniques: a comprehensive survey,” Multimedia Systems, vol. 24,
no. 2, pp. 211–240, Mar. 2018, doi: 10.1007/s00530-017-0538-9.
[5] H. Yao, S. Wang, Y. Zhao, and X. Zhang, “Detecting image forgery using perspective constraints,” IEEE Signal Processing
Letters, vol. 19, no. 3, pp. 123–126, Mar. 2012, doi: 10.1109/LSP.2011.2182191.
[6] M. Barni and A. Costanzo, “A fuzzy approach to deal with uncertainty in image forensics,” Signal Processing: Image
Communication, vol. 27, no. 9, pp. 998–1010, Oct. 2012, doi: 10.1016/j.image.2012.07.006.
[7] D. V. -Padin and F. P. -Gonzalez, “Prefilter design for forensic resampling estimation,” in 2011 IEEE International Workshop on
Information Forensics and Security, IEEE, Nov. 2011, pp. 1–6, doi: 10.1109/WIFS.2011.6123133.

Int J Artif Intell, Vol. 14, No. 2, April 2025: 1471-1483

Int J Artif Intell ISSN: 2252-8938  1481

[8] J. Zhang, K. Yu, Z. Wen, X. Qi, and A. Kumar Paul, “3D reconstruction for motion blurred images using deep learning-based
intelligent systems,” Computers, Materials & Continua, vol. 66, no. 2, pp. 2087–2104, 2021, doi: 10.32604/cmc.2020.014220.
[9] A. Lattas et al., “AvatarMe: realistically renderable 3d facial reconstruction ‘in-the-wild,’” in IEEE/CVF Conference on
Computer Vision and Pattern Recognition, CVPR 2020, 2020, pp. 760–769.
[10] T. Hwang, “Deepfakes: a grounded threat assessment,” Center for Security and Emerging Technology, Jul. 2020, doi:
10.51593/20190030.
[11] S. Lyu, “Deepfake detection: current challenges and next steps,” in 2020 IEEE International Conference on Multimedia & Expo
Workshops (ICMEW), IEEE, Jul. 2020, pp. 1–6, doi: 10.1109/ICMEW46912.2020.9105991.
[12] C. Yang, L. Ding, Y. Chen, and H. Li, “Defending against gan-based deepfake attacks via transformation-aware adversarial
faces,” in 2021 International Joint Conference on Neural Networks (IJCNN), IEEE, Jul. 2021, pp. 1–8, doi:
10.1109/IJCNN52387.2021.9533868.
[13] X. Liu et al., “Self-supervised learning: generative or contrastive,” IEEE Transactions on Knowledge and Data Engineering,
vol. 35, no. 1, pp. 857-876, 1 Jan. 2023, doi: 10.1109/TKDE.2021.3090866
[14] W. Wang, Q. Huang, S. You, C. Yang, and U. Neumann, “Shape inpainting using 3d generative adversarial network and recurrent
convolutional networks,” in 2017 IEEE International Conference on Computer Vision (ICCV), IEEE, Oct. 2017, pp. 2317–2325,
doi: 10.1109/ICCV.2017.252.
[15] D.-N. Zhao, R.-K. Wang, and Z.-M. Lu, “Inter-frame passive-blind forgery detection for video shot based on similarity analysis,”
Multimedia Tools and Applications, vol. 77, no. 19, pp. 25389–25408, Oct. 2018, doi: 10.1007/s11042-018-5791-1.
[16] J. A. Aghamaleki and A. Behrad, “Inter-frame video forgery detection and localization using intrinsic effects of double
compression on quantization errors of video coding,” Signal Processing: Image Communication, vol. 47, pp. 289–302, Sep. 2016,
doi: 10.1016/j.image.2016.07.001.
[17] C. Long, A. Basharat, and A. Hoogs, “A coarse-to-fine deep convolutional neural network framework for frame duplication
detection and localization in forged videos,” in IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)
Workshops, CVPR 2019, 2019, pp. 1–10.
[18] Y. Q. Shi, C. Chen, and W. Chen, “A natural image model approach to splicing detection,” in Proceedings of the 9th workshop on
Multimedia & security - MM&Sec ’07, New York, USA: ACM Press, 2007, doi: 10.1145/1288869.1288878.
[19] Y. Deng, J. Yang, D. Chen, F. Wen, and X. Tong, “Disentangled and controllable face image generation via 3d imitative-
contrastive learning,” in 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), IEEE, Jun. 2020,
pp. 5153–5162, doi: 10.1109/CVPR42600.2020.00520.
[20] J. Zhao, M. Mathieu, and Y. LeCun, “Energy-based generative adversarial network,” MM 2017 - Proceedings of the 2017 ACM
Multimedia Conference, pp. 672–680, Sep. 2016, doi: 10.1145/3123266.3123334.
[21] S. A. Buo, “The emerging threats of deepfake attacks and countermeasures,” arXiv-Computer Science, pp. 1-5, 2020.
[22] L. Verdoliva, “Media forensics and deepfakes: an overview,” IEEE Journal of Selected Topics in Signal Processing, vol. 14,
no. 5, pp. 910–932, Aug. 2020, doi: 10.1109/JSTSP.2020.3002101.
[23] L. Guarnera, O. Giudice, M. Niesner, and S. Battiato, “On the exploitation of deepfake model recognition,” in 2022 IEEE/CVF
Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), IEEE, Jun. 2022, pp. 61–70, doi:
10.1109/CVPRW56347.2022.00016.
[24] G. K. Birajdar and V. H. Mankar, “Digital image forgery detection using passive techniques: a survey,” Digital Investigation,
vol. 10, no. 3, pp. 226–245, 2013, doi: 10.1016/j.diin.2013.04.007.
[25] A. Singh, A. S. Saimbhi, N. Singh, and M. Mittal, “Deepfake video detection: a time-distributed approach,” SN Computer
Science, vol. 1, no. 4, p. 212, Jul. 2020, doi: 10.1007/s42979-020-00225-9.
[26] P. Charitidis, G. Kordopatis-Zilos, S. Papadopoulos, and I. Kompatsiaris, “Investigating the impact of pre-processing and
prediction aggregation on the deepfake detection task,” arXiv-Computer Science, pp. 1-11, 2020.
[27] C. Öhman, “Introducing the pervert’s dilemma: a contribution to the critique of deepfake pornography,” Ethics and Information
Technology, vol. 22, no. 2, pp. 133–140, Jun. 2020, doi: 10.1007/s10676-019-09522-1.
[28] A. Khodabakhsh, R. Ramachandra, K. Raja, P. Wasnik, and C. Busch, “Fake face detection methods: can they be generalized?,”
in 2018 International Conference of the Biometrics Special Interest Group (BIOSIG), IEEE, Sep. 2018, pp. 1–6, doi:
10.23919/BIOSIG.2018.8553251.
[29] A. Rossler, D. Cozzolino, L. Verdoliva, C. Riess, J. Thies, and M. Niessner, “Faceforensics++: learning to detect manipulated
facial images,” in 2019 IEEE/CVF International Conference on Computer Vision (ICCV), IEEE, Oct. 2019, pp. 1–11, doi:
10.1109/ICCV.2019.00009.
[30] A. Buslaev, V. I. Iglovikov, E. Khvedchenya, A. Parinov, M. Druzhinin, and A. A. Kalinin, “Albumentations: fast and flexible
image augmentations,” Information, vol. 11, no. 2, 2020, doi: 10.3390/info11020125.
[31] M. K. Hassan, M. R. Hassan, M. M. T. Ahmed, M. S. A. Sabbir, M. S. Ahmed, and M. Biswas, “A survey on an intelligent
system for persons with visual disabilities,” Australian Journal of Engineering and Innovative Technology, vol. 3, no. 6,
pp. 97–118, Nov. 2021, doi: 10.34104/ajeit.021.0970118.
[32] I. Amerini and R. Caldelli, “Exploiting prediction error inconsistencies through lstm-based classifiers to detect deepfake videos,”
in Proceedings of the 2020 ACM Workshop on Information Hiding and Multimedia Security, New York, USA: ACM, Jun. 2020,
pp. 97–102, doi: 10.1145/3369412.3395070.
[33] J. Lin, Y. Li, and G. Yang, “FPGAN: face de-identification method with generative adversarial networks for social robots,”
Neural Networks, vol. 133, pp. 132–147, Jan. 2021, doi: 10.1016/j.neunet.2020.09.001.
[34] Benpflaum et al., “Deepfake detection challenge,” Kaggle. 2019. Accessed: Mar. 01, 2022. [Online]. Available:
https://kaggle.com/competitions/deepfake-detection-challenge
[35] L. D’Amiano, D. Cozzolino, G. Poggi, and L. Verdoliva, “Video forgery detection and localization based on 3d patchmatch,” in
2015 IEEE International Conference on Multimedia & Expo Workshops (ICMEW), IEEE, Jun. 2015, pp. 1–6, doi:
10.1109/ICMEW.2015.7169805.
[36] C.-H. Choi, M.-J. Lee, and H.-K. Lee, “Scanner identification using spectral noise in the frequency domain,” in 2010 IEEE
International Conference on Image Processing, IEEE, Sep. 2010, pp. 2121–2124, doi: 10.1109/ICIP.2010.5652108.
[37] T. Park, M.-Y. Liu, T.-C. Wang, and J.-Y. Zhu, “Semantic image synthesis with spatially-adaptive normalization,” in 2019
IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), IEEE, Jun. 2019, pp. 2332–2341, doi:
10.1109/CVPR.2019.00244.
[38] Y. Nirkin, Y. Keller, and T. Hassner, “FSGAN: subject agnostic face swapping and reenactment,” in 2019 IEEE/CVF
International Conference on Computer Vision (ICCV), IEEE, Oct. 2019, pp. 7183–7192, doi: 10.1109/ICCV.2019.00728.
[39] DARPA, “MediFor: media forensics,” Defense Advanced Research Projects Agency, 2020. [Online]. Available:
https://www.darpa.mil/program/media-forensics
Video forgery: An extensive analysis of inter-and intra-frame manipulation alongside … (Sumaiya Shaikh)
1482  ISSN: 2252-8938

[40] J. Thies, M. Elgharib, A. Tewari, C. Theobalt, and M. Nießner, “Neural voice puppetry: audio-driven facial reenactment,” in
Computer Vision – ECCV 2020: 16th European Conference, 2020, pp. 716–731, doi: 10.1007/978-3-030-58517-4_42.
[41] P. Gupta, K. Chugh, A. Dhall, and R. Subramanian, “The eyes know it,” in Proceedings of the 2020 International Conference on
Multimodal Interaction, New York, USA: ACM, Oct. 2020, pp. 519–527, doi: 10.1145/3382507.3418857.
[42] L. Li, J. Bao, H. Yang, D. Chen, and F. Wen, “Advancing high fidelity identity swapping for forgery detection,” in 2020
IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), IEEE, Jun. 2020, pp. 5073–5082, doi:
10.1109/CVPR42600.2020.00512.
[43] L. Chai, D. Bau, S.-N. Lim, and P. Isola, “What makes fake images detectable? understanding properties that generalize,” in
Computer Vision – ECCV 2020: 16th European Conference, 2020, pp. 103–120, doi: 10.1007/978-3-030-58574-7_7.
[44] X. Chang, J. Wu, T. Yang, and G. Feng, “Deepfake face image detection based on improved vgg convolutional neural network,”
in 2020 39th Chinese Control Conference (CCC), IEEE, Jul. 2020, pp. 7252–7256, doi: 10.23919/CCC50068.2020.9189596.
[45] Y. Li, X. Yang, P. Sun, H. Qi, and S. Lyu, “Celeb-df: a large-scale challenging dataset for deepfake forensics,” in 2020
IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), IEEE, Jun. 2020, pp. 3204–3213, doi:
10.1109/CVPR42600.2020.00327.
[46] T. Bianchi and A. Piva, “Detection of nonaligned double jpeg compression based on integer periodicity maps,” IEEE
Transactions on Information Forensics and Security, vol. 7, no. 2, pp. 842–848, Apr. 2012, doi: 10.1109/TIFS.2011.2170836.
[47] M. S. Rana, M. N. Nobi, B. Murali, and A. H. Sung, “Deepfake detection: a systematic literature review,” IEEE Access, vol. 10,
pp. 25494–25513, 2022, doi: 10.1109/ACCESS.2022.3154404.
[48] Z. Boulkenafet, J. Komulainen, and A. Hadid, “Face anti-spoofing based on color texture analysis,” in 2015 IEEE International
Conference on Image Processing (ICIP), IEEE, Sep. 2015, pp. 2636–2640, doi: 10.1109/ICIP.2015.7351280.
[49] Z. Zhang, J. Hou, Z. Li, and D. Li, “Inter-frame forgery detection for static-background video based on mvp consistency,” in
International Workshop on Digital Watermarking, Springer, Cham, 2016, pp. 94–106, doi: 10.1007/978-3-319-31960-5_9.
[50] S. Hussain, P. Neekhara, M. Jere, F. Koushanfar, and J. McAuley, “Adversarial deepfakes: evaluating vulnerability of deepfake
detectors to adversarial examples,” in 2021 IEEE Winter Conference on Applications of Computer Vision (WACV), IEEE, Jan.
2021, pp. 3347–3356, doi: 10.1109/WACV48630.2021.00339.
[51] T. Zhao, X. Xu, M. Xu, H. Ding, Y. Xiong, and W. Xia, “Learning self-consistency for deepfake detection,” in 2021 IEEE/CVF
International Conference on Computer Vision (ICCV), IEEE, Oct. 2021, pp. 15003–15013, doi: 10.1109/ICCV48922.2021.01475.
[52] H. H. Nguyen, F. Fang, J. Yamagishi, and I. Echizen, “Multi-task learning for detecting and segmenting manipulated facial
images and videos,” in 2019 IEEE 10th International Conference on Biometrics Theory, Applications and Systems (BTAS), IEEE,
Sep. 2019, pp. 1–8, doi: 10.1109/BTAS46853.2019.9185974.
[53] F. Chollet, “Xception: deep learning with depthwise separable convolutions,” 30th IEEE Conference on Computer Vision and
Pattern Recognition, CVPR 2017, pp. 1800–1807, 2017, doi: 10.1109/CVPR.2017.195.
[54] D. A. Coccomini, N. Messina, C. Gennaro, and F. Falchi, “Combining efficientnet and vision transformers for video deepfake
detection,” in International Conference on Image Analysis and Processing, Springer, Cham, 2022, pp. 219–229, doi:
10.1007/978-3-031-06433-3_19.
[55] F. Lago, C. Pasquini, R. Bohme, H. Dumont, V. Goffaux, and G. Boato, “More real than real: a study on human visual perception
of synthetic faces [applications corner],” IEEE Signal Processing Magazine, vol. 39, no. 1, pp. 109–116, Jan. 2022, doi:
10.1109/MSP.2021.3120982.
[56] T. Dzanic, K. Shah, and F. D. Witherden, “Fourier spectrum discrepancies in deep network generated images,” in Proceedings of
the 34th International Conference on Neural Information Processing Systems, Curran Associates Inc., 2020, pp. 3022–3032.
[57] K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,” Proceedings of the IEEE Computer Society
Conference on Computer Vision and Pattern Recognition, vol. 2016-Decem, pp. 770–778, 2016, doi: 10.1109/CVPR.2016.90.
[58] A. Swaminathan, M. Wu, and K. J. R. Liu, “Digital image forensics via intrinsic fingerprints,” IEEE Transactions on Information
Forensics and Security, vol. 3, no. 1, pp. 101–117, Mar. 2008, doi: 10.1109/TIFS.2007.916010.
[59] M. Goljan, J. Fridrich, and T. Filler, “Managing a large database of camera fingerprints,” in Media Forensics and Security II, part
of the IS&T-SPIE Electronic Imaging Symposium, Feb. 2010, p. 754108, doi: 10.1117/12.838378.
[60] S. Bayram, H. Sencar, N. Memon, and I. Avcibas, “Source camera identification based on cfa interpolation,” in IEEE
International Conference on Image Processing 2005, IEEE, 2005, doi: 10.1109/ICIP.2005.1530330.
[61] O. Giudice, A. Paratore, M. Moltisanti, and S. Battiato, “A classification engine for image ballistics of social data,” in International
Conference on Image Analysis and Processing, Springer, Cham, 2017, pp. 625–636, doi: 10.1007/978-3-319-68548-9_57.
[62] Z. He, W. Zuo, M. Kan, S. Shan, and X. Chen, “AttGAN: facial attribute editing by only changing what you want,” IEEE
Transactions on Image Processing, vol. 28, no. 11, pp. 5464–5478, Nov. 2019, doi: 10.1109/TIP.2019.2916751.
[63] H. Mo, B. Chen, and W. Luo, “Fake faces identification via convolutional neural network,” in Proceedings of the 6th ACM
Workshop on Information Hiding and Multimedia Security, ACM, Jun. 2018, pp. 43–47, doi: 10.1145/3206004.3206009.
[64] L. Jiang, R. Li, W. Wu, C. Qian, and C. C. Loy, “DeeperForensics-1.0: a large-scale dataset for real-world face forgery
detection,” in 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), IEEE, Jun. 2020,
pp. 2886–2895, doi: 10.1109/CVPR42600.2020.00296.
[65] Y.-J. Heo, Y.-J. Choi, Y.-W. Lee, and B.-G. Kim, “Deepfake detection scheme based on vision transformer and distillation,”
arXiv-Computer Science, pp. 1-7, Apr. 2021.
[66] M. T. Jafar, M. Ababneh, M. Al-Zoube, and A. Elhassan, “Forensics and analysis of deepfake videos,” in 2020 11th International
Conference on Information and Communication Systems (ICICS), IEEE, Apr. 2020, pp. 53–58, doi:
10.1109/ICICS49469.2020.239493.
[67] B. Dolhansky et al., “The deepfake detection challenge (DFDC) dataset,” arXiv-Computer Science, pp. 1-13, Jun. 2020.
[68] R. Chesney and D. Citron, “Deepfakes and the new disinformation war-the coming age of post-truth geopolitics,” Foreign
Relations, Inc. 2019. [Online]. Available: https://www.foreignaffairs.com/articles/world/2018-12-11/deepfakes-and-new-
disinformation-war
[69] T. Mittal, U. Bhattacharya, R. Chandra, A. Bera, and D. Manocha, “Emotions don’t lie,” in Proceedings of the 28th ACM
International Conference on Multimedia, New York, USA: ACM, Oct. 2020, pp. 2823–2832, doi: 10.1145/3394171.3413570.
[70] P. Zhou, X. Han, V. I. Morariu, and L. S. Davis, “Two-stream neural networks for tampered face detection,” in 2017 IEEE
Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), IEEE, Jul. 2017, pp. 1831–1839, doi:
10.1109/CVPRW.2017.229.
[71] E. Sabir, J. Cheng, A. Jaiswal, W. AbdAlmageed, I. Masi, and P. Natarajan, “Recurrent convolutional strategies for face
manipulation detection in videos,” in IEEE Conference on Computer Vision and Pattern Recognition Workshops, CVPR 2019,
2019, pp. 80–87.

Int J Artif Intell, Vol. 14, No. 2, April 2025: 1471-1483

Int J Artif Intell ISSN: 2252-8938  1483

[72] R. Tolosana, R. Vera-Rodriguez, J. Fierrez, A. Morales, and J. Ortega-Garcia, “Deepfakes and beyond: a survey of face
manipulation and fake detection,” Information Fusion, vol. 64, pp. 131–148, Dec. 2020, doi: 10.1016/j.inffus.2020.06.014.
[73] W. Ahmed, F. Shahzad, A. R. Javed, F. Iqbal, and L. Ali, “WhatsApp network forensics: discovering the ip addresses of
suspects,” 2021 11th IFIP International Conference on New Technologies, Mobility and Security, NTMS 2021, 2021, doi:
10.1109/NTMS49979.2021.9432677.

BIOGRAPHIES OF AUTHORS

Sumaiya Shaikh received her B.E. degree in electronics communication and

engineering from SRK Institute of Technology, Andhra Pradesh, India, in 2013. She has
received her M.E. degree in computer science and engineering from V. R. Siddhartha
Engineering College, Vijayawada, Andhra Pradesh, India, in 2018. Her research interest
includes cyber security, cryptography, and digital forensics. She has published more than 20
papers in reputed international conferences/journals. A Research Scholar, Department of
Computer Science and Engineering, Koneru Lakshmaiah Education Foundation,
Vaddeswaram, Guntur, Andhra Pradesh, India. She can be contacted at email:
[email protected].

Sathish Kumar Kannaiah received his B.E. degree in computer science and
engineering from Madurai Kamaraj University, Tamilnadu, India, in the year 2002. He has
received his M.E. degree in computer science and engineering from Anna University
Tiruchirappalli, Tamilnadu, India, in the year 2010. He received his Ph.D. degree in Anna
University Chennai in the year 2020. His research interest includes cryptography and network
security, and he has published more than 20 papers in reputed international
conferences/journals. Currently working as Associate Professor, Department of Computer
Science and Engineering, Koneru Lakshmaiah Education Foundation, Vaddeswaram, Guntur,
Andhra Pradesh, India. He can be contacted at email: [email protected].

Video forgery: An extensive analysis of inter-and intra-frame manipulation alongside … (Sumaiya Shaikh)