Published by : International Journal of Engineering Research & Technology (IJERT)

http://www.ijert.org ISSN: 2278-0181

Vol. 11 Issue 01, January-2022

Real Time Object Detection and Recognition

using MobileNet-SSD with OpenCV
Mr. Harshal Honmote Mr. Pranav Katta
Computer Engineering Computer Engineering
BSIOTR, Pune BSIOTR, Pune

Mr. Shreyas Gadekar Prof. Madhavi Kulkarni

Computer Engineering Assistant Professor
BSIOTR, Pune BSIOTR, Pune

Abstract-- Real time object detection is an immense, vibrant Precision (AP) of the algorithm to recognize various classes
and complex area of computer vision. Assuming there is a as vehicle, person and chair is 99.76%, 97.76% and 71.07%,
single object to be distinguished in an image, it is known as separately. This improves the accuracy of behavior detection
Image Localization and in the event that there are various at a handling speed which is needed for the real-time
objects in an image, then, at that point, it is Object Detection.
Mobile networks and binary neural networks are the most
location and the necessities of day by day observing indoor
generally involved techniques for current deep learning models and outside. The mix of MobileNet into the SSD framework
to perform different tasks on embedded systems. In this paper, forms one of the center parts of our work.
we develop a method to distinguish an item thinking about the However, MobileNet with the effective SSD framework has
deep learning pre-prepared model MobileNet for Single Shot been a hot exploration point in recent times, to a great extent
Multi-Box Detector (SSD). This algorithm is used for real-time because of managing the functional limits of running strong
detection and for webcam streaming to detect object in a video neural nets on low-end devices like cell phones/laptops to
stream. Subsequently, we utilize an object detection module additionally expand the horde of conceivable outcomes with
that can identify what is in the video stream. To carry out the respect to real-time applications.
module, we join the MobileNet and the SSD framework for a
quick and efficient deep learning-based strategy for object
identification. II. LITERATURE SURVEY
A. MobileNet-SSD
Keywords—MobileNet, SSD (Single Shot Multi-Box Detector).
Our proposed model depends on the MobileNet-
SSD architecture. One reason why we chose this architecture
I. INTRODUCTION is on the because that as shown in the paper [2], it gives good
Object detection is one of the most important fields object detection accuracy while being quicker than different
of exploration in computer vision today. It is an architectures, for example, YOLO. Especially, this is valid
augmentation of image classification the objective is to when attempting to detect object in real time in low
identify one or more classes of objects in a picture and with computing devices as in our system. MobileNet-SSD
the help of bounding boxes locate their presence. permits to lessen the detection time by addressing the model
Consequently, object detection carries an important role in utilizing 8-bit integers rather than 32-bit floats. The input of
many real-world applications like image recovery and the model was set to an image with 300 by 300 pixels and
video surveillance. the result of the model addressed the position of the
The main purpose of our analysis is to elaborate the bounding box as well as the detection confidences (from 0
accuracy of an object detection technique SSD and the pre- to 1) for each identified object. A detection confidence
trained deep learning model MobileNet and additionally threshold of 0.5 was utilized to decide if the detected object
feature a portion of the notable elements that make this was valid.[3]
method stand out. The trial results show that the Average

Fig. 1. SSD-based detection with MobileNet as backbone.[1]

IJERTV11IS010070 www.ijert.org 141

(This work is licensed under a Creative Commons Attribution 4.0 International License.)
Published by : International Journal of Engineering Research & Technology (IJERT)
http://www.ijert.org ISSN: 2278-0181
Vol. 11 Issue 01, January-2022

B. OpenCV (Open-Source computer vision) IV. CONCLUSION

OpenCV is a library of programming functions In this research, we proposed a deep learning model
basically focused on real time computer vision. OpenCV is to identify progressively the place of the object in pictures.
an open-source library which is useful for computer vision The framework could distinguish the item with a normal
applications like CCTV film analysis, video analysis and accuracy like other best in class frameworks. In this way, we
image analysis. It is an incredible tool for image processing utilize an object detection module that can recognize what is
and performing computer vision tasks. OpenCV is written in the real time video stream. To carry out the module, we
by C++ and has in excess of 2,500 optimized algorithms. [5] join the MobileNet and the SSD framework for a quick and
At the point when we make applications for computer vision productive deep learning-based strategy for object detection.
that we do not want to make it from scratch instead we can In future work, we will keep on enhancing our detection
utilize this library to begin focusing on real world problems. network model, including lessening memory utilization and
OpenCV has a function to read video, which is speeding up and additionally we will add more classes.
cv2.VideoCapture(). We can access webcam by passing 0 as
function parameter. To catch CCTV film then we can pass V. REFERENCES
RTSP URL in the function parameter, which is truly [1] Yundong Zhang, Haomin Peng haomin and Pan Hu, “Towards Real-
time Detection and Camera Triggering,” CS341.
valuable for video analysis.
[2] Ibai Gorordo Fernandez and Chikamune Wada, “Shoe Detection Using
SSD-MobileNet Architecture,”2020 IEEE 2nd Global Conference on
III. PROPOSED SYSTEM Life Sciences and Technologies (LifeTech 2020).
In the Proposed System, we are going to detect [3] Yu-Chen Chiu, Chi-Yi Tsai, Mind-Da Ruan, Guan-Yu Shen and Tsu-
Tian Lee, “Mobilenet-SSDv2: An Improved Object Detection Model for
objects in real time with the help of Mobilenet-SSD model
Embedded Systems,” ©2020 IEEE.
in fast and efficient way. We will create the Python script for [4] Andres Heredia and Gabriel Barros-Gavilanes,” Video processing inside
object detection using deep neural network with OpenCV embedded devices using SSD-Mobilenet to count mobility actors,” 978-
3.4. 1-7281-1614-3/19 ©2019 IEEE.
[5] G. Bradski and, A. Kaehler, “Learning OpenCV”, OReilly Publications,
Working of the system is as follow:
2008.
Input will be given through Realtime video by [6] Animesh Srivastava1, Anuj Dalvi2, Cyrus Britto3, Harshit Rai4, Kavita
camera or webcam, based on streamlined MobileNet Shelke5,” Explicit Content Detection using Faster R-CNN and SSD
Architecture which uses depth-wise separable convolutions MobileNet v2,” e-ISSN: 2395-0056 © 2020, IRJET.
[7] R. Huang, J. Pedoeem, and C. Chen, “YOLO-LITE: A Real-Time Object
to build light weight deep neural Networks. The input video
Detection Algorithm Optimized for Non-GPU Computers,” in
divided into frames and pass it to MobileNet layers. [4] Each Proceedings - 2018 IEEE International Conference on Big Data, Big
feature value is determined as a difference between the Data 2018.
amount of pixel intensity under the bright region and the
pixel intensity under the dark area. Every one of the possible
sizes and area of the image is utilized to compute these
elements. An image may contain irrelevant features and few
relevant characteristics that can be used to detect the object.
The job of the MobileNet layers is to change over
the pixels from the input image into highlights that describe
the contents of the image. Then it passes to MobileNet-SSD
model to determine the bounding boxes and corresponding
class (label) of objects. After that the only last step is to show
or display the Output.

PROPOSED SYSTEM ARCHHITECTURE DIAGRAM:

Fig.2. Proposed System Architecture Diagram

IJERTV11IS010070 www.ijert.org 142

(This work is licensed under a Creative Commons Attribution 4.0 International License.)