COURSE: Advanced VLSI Design (MEE 1123)

ASSIGNMENT (CLO 5, PLO 7, C6, A4): Prepare an assignment in CMOS fabrication

process. You may explain your answer by using an appropriate equation and diagram.

Rubric for the report of assignment

Poor Moderate Good Excellent
Criteria
(1) (2) (3) (4)
Introductory Introductory Introductory Introductory
paragraph is not paragraph is clearly paragraph is clearly paragraph is very
apparent. stated with a focus. stated, focused, well- clearly stated,
Introduction
written and impacts distinct focus, well-
the report. written and enhances
the impact of report.
The topic has not The topic is The topic discussed The topic discussed
been clearly discussed and and elaborated clearly specifically and well
discussed elaborates generally together with relevant researched
Content
example. documented topics
for discussion with
relevant examples.
Do not have any or Contained figures / Contained relevant Contained relevant
related figures / attachments to figures / attachments and well manage
Figures/
attachments to support the report to support the report figures / attachments
Attachments
support the report description but did description. to support the report
description. not relevant. description.
Concluding Concluding Concluding paragraph Concluding
paragraph is not paragraph is only follows and paragraph draws a
Conclusion/ clearly apparent/ remotely related to summarizes report clear and effective
Suggestion suggestion. the report discussion discussion. Have 2 – 3 conclusion report.
& 1 rational suggestions Have > 4
suggestion. suggestion.
Used poor English Moderate in English Good in English Excellent in English
Languages &
Language writing Language writing Language writing Language writing
Neatness
skill. skill. skill. skill.
Does not follow all Follow 2 to 3 Follow ALL formats Follow ALL formats
the formats applied formats for the for the report exactly. for the report exactly
Format
for the report report properly. and complete reports
writing.

Prepared By: Dr. Noraisyah Tajudin.

 FACULTY OF ENGINEERING AND BUILT
ENVIRONMENT

MASTER OF SCIENCE IN ELECTRICALS AND

ELECTRONICS ENGINEERING & MASTER OF
SCIENCE IN TELECOMMUNICATION
ENGINEERING.

IMAGE PROCESSING
MET 1233

ASSIGNMENT
Prepare the full report regarding the below issues:
In the modern world, companies and organizations around the world use AI image
processing. AI image processing, also known as AI-powered image processing or computer
vision, refers to the use of artificial intelligence (AI) techniques to analyze, interpret, and
manipulate digital images and videos. It involves developing algorithms and models that
enable computers to understand and extract information from visual data, similar to how
humans perceive and comprehend images. AI image processing combines computer science,
machine learning, and signal processing to perform a variety of tasks on images, such as;
character recognition, data extraction, pattern recognition, visualization, identification, and
classification, among others.
Prepare a written assignment on two (2) AI image processing techniques using online
and free tools. Use your own images that you've taken using a camera or a smartphone.
Report:
The softcopy of your report must contain the following:
 Introduction
 AI image processing (at least 2 software).
 Image processing techniques that are used to enhance your images. Each AI software
must be at least showing three (3) Image processing techniques.
 Briefly explain the image processing techniques that are involved in your image
enhancement etc.
 Conclusion
 References

*Your full report must be exceeded than 4 pages

Final submission of assignment report:
12th September 2023 (Tuesday)
Report on AI Image Processing Techniques

1. Introduction

AI image processing, also referred to as computer vision, is a revolutionary discipline that precisely
analyses, interprets, and manipulates pictures using machine learning, computer technology, including
signal processing. Social networking and medical diagnostics are only two of the businesses that have
used it. AI image processing uses signal processing, computer science, and machine learning methods
to comprehend pictures in a way that is comparable to that of humans. AI systems can identify
objects, decipher patterns, and glean insightful information from photos and videos by training on vast
datasets. Canva and Adobe Photoshop Express are two widely available, free apps that are now
pushing AI image processing. Adobe Photoshop Express, which offers tools like auto-enhance,
elimination of background noise reduction, including sharpness tweaks, uses advanced AI models to
assist users in rapidly improving picture quality. With tools like Display-colour improvement, auto-
cropping, and sharpness changes, Canva, a well-known design platform, uses AI to assist users
improve picture composition and clarity. AI processing of pictures has important practical
applications in a variety of sectors, including retail, healthcare, and agriculture. AI systems may help
radiologists in the medical field by pointing out anomalies in MRIs or X-rays, improving the precision
of diagnoses. AI-powered image identification in retail streamlines the shopping experience by
enabling consumers to look for items simply snapping a photo. AI can evaluate satellite photos in
agriculture to better manage resources and track crop health. More contextually conscious algorithms
that comprehend pictures within particular cultural or environmental settings, improving their
accuracy and relevance, are among the future possibilities for AI image processing. These
developments imply that AI processing images will continue to be a crucial technology in a variety of
fields, including marketing, journalism, public safety, and transportation. Innovations in 3D stacking,
AI-driven chip design, 6nm and 5nm manufacturing nodes, and elsewhere at the interface of AI and
VLSI are driving ongoing change. Keep yourself informed about these breakthroughs and be open to
new tools and approaches if you want to push the limits of what an AI-optimized VLSI can do.
Specific to the application Integrated Circuits (ASICs) alongside AI Accelerators in Very Large Scale
Integration (VLSI), Neuromorphic Computing, Technology Design Automation (EDA) instruments
Hardware Description The languages (HDLs), Authority Efficiency alongside Thermal Management
to feed AI Chips, Artificial Intelligence in Field Programmable Gate Array (FPGA)-based Very Large
Scale Integration (VLSI) Systems, Optimizations of Algorithms for Very Large Scale Integration
(VLSI-Driven AI), Coding for Distributed along with Heterogeneous Because of their efficiency in
parallel operations and optimization for machine learning tasks, ASICs are crucial for improving AI
processing capabilities. Some examples of very large scale integration (VLSI) implementations that
focus on neural network processing operations include the Tensor Processing Units (TPUs) from
Google and the Deep Learning Accelerator (NVDLA) from NVIDIA. Lodhi, an Intel neuromorphic
VLSI chip, is one example of a device that combines thousands of neurons and synapses to provide
AI applications at the edge that are energy efficient. Virtual logic system (VLSI) design optimization
requires electronic design automation (EDA) tools, particularly AI-powered ones, for tasks like
placement, navigation, and timing analysis. Artificial intelligence systems may save design time and
improve accuracy by predicting the ideal circuit routes. By automating the process of chip layout
design, tools such as Cadence Cerberus along with Synopsys DSO.ai use reinforcement learning to
achieve better optimization of chip performance as well as efficiency compared to more conventional
approaches. Particularly in the area of digital logic synthesis that can effectively handle matrix
multiplications as well as vector operations essential to deep learning, advanced coding approaches in
HDLs like System Verilog and VHDL have included additional AI capabilities. Specialized very large
scale integration (VLSI) AI chips use HDLs to design AI-optimized hardware components, such as
Matrix Multiplication components (MMUs) as well as Processing Elements (PEs). Deploying AI
processors in energy-constrained contexts, such as the Internet of Things (IoT) or mobile devices,
requires careful consideration of power efficiency including thermal control. For AI models to work
with VLSI design, optimization of algorithms is essential, especially for quantization and pruning
approaches. This optimization allows for low-power, high-speed AI on VLSI processors. By the end
of the day, AI applications—especially in safety-critical systems involving medical devices and
automotive—need fault-tolerant and dependable VLSI architectures. Deep neural networks (DNNs)
with built-in redundancy and hardware-software co-design the fact that can identify and fix errors in
real-time without affecting the entire system are examples of advanced AI coding in this sector.
AI processing photographs is a prime example of how technology may revolutionized both the
personal and professional spheres by providing individuals with unprecedented capabilities for
creating, analyzing, and comprehending visual data. The emergence of AI-powered programs like
Canva and Adobe Photoshop Express is just the first step; AI image processing will alter our
interactions with pictures in a myriad of ways in the future.

`2. AI IMAGE PROCESSING TOOLS

This report focuses on two widely accessible, free AI image processing tools: Adobe
Photoshop Express and Canva. Each tool has unique capabilities and is equipped with a suite
of AI-based features designed to enhance image quality and appearance.

2.1 “ADOBE PHOTOSHOP EXPRESS”

Adobe Photoshop Express is a popular, free-to-use mobile and desktop application that offers
various AI-powered features for image enhancement. Its user-friendly interface allows for
complex image manipulation with ease. Photoshop Express utilizes AI algorithms to provide
a range of effects and filters designed to optimize image quality.

A. Image Processing Techniques Used In Adobe Photoshop Express:

Auto-Enhance (AI-Powered Image Optimization): This feature uses AI to
automatically adjust brightness, contrast, and exposure to improve the overall
appearance of an image. By analyzing image characteristics, the AI enhances colors
and clarity without user input, making it ideal for quick adjustments.
Background Removal (Segmentation): Adobe Photoshop Express has a background
removal tool that uses AI segmentation techniques to isolate the subject of an image
 from the background. This allows users to remove or replace backgrounds seamlessly,
making the image’s subject more prominent.
Noise Reduction (Image Denoising): AI-based noise reduction techniques in
Photoshop Express allow for the removal of grainy or pixelated artifacts, especially in
low-light images. The tool analyzes the pixel structure of the image and applies an
algorithm to smooth out imperfections while preserving details.

2.2 “CANVA”

Canva is a widely-used graphic design platform that incorporates AI-based image processing
tools suitable for both novice and experienced users. It is accessible through mobile and
desktop applications, and it provides a variety of features to enhance visual appeal.

Image Processing Techniques Used in Canva: Image Sharpness and Clarity

Adjustment: Canva’s AI algorithms adjust image sharpness and clarity, automatically
refining edges and enhancing details to produce a crisper image. This technique is
especially useful for portrait or close-up shots.
Color Enhancement and Filter Application: The tool’s color enhancement feature
uses AI to analyze image tones and apply adjustments that enhance color vibrancy
and contrast. Canva also offers a variety of filters that can alter the mood or aesthetics
of an image while maintaining clarity and detail.
Auto-Alignment and Cropping (Smart Cropping): Canva’s auto-alignment feature uses AI to
analyze the focal points within an image, automatically cropping and aligning elements for
balanced composition. This feature helps in resizing or framing images without
compromising their visual integrity.

3. IMAGE PROCESSING TECHNIQUES EXPLAINED

Deeper look at how each technique works to enhance images and why it’s useful for
improving visual quality:

Auto-Enhance in Adobe Photoshop Express: This feature employs AI to analyze

the image’s attributes, such as brightness, contrast, and saturation. Based on its
analysis, it applies adjustments that create a balanced exposure and natural color
tones, often with a single tap.
Background Removal in Adobe Photoshop Express: This technique, powered by
AI segmentation, detects and isolates the main subject from its background, allowing
users to either replace or remove the background entirely. This technique is
particularly beneficial for product photography and portrait images, making subjects
stand out more effectively.
Noise Reduction in Adobe Photoshop Express: The noise reduction tool identifies
unwanted grain or noise patterns in an image and smooths them out, preserving
important details. This is especially effective for images captured in low-light
environments, where digital noise tends to appear more prominently.
Image Sharpness and Clarity Adjustment in Canva: Using edge detection
algorithms, this feature sharpens edges and enhances the details in an image. By
increasing sharpness, it improves overall clarity, which can be useful for portrait or
close-up images.
Color Enhancement and Filter Application in Canva: This feature utilizes AI to
detect underrepresented colors in the image and enhances their vibrancy, creating a
more visually appealing image. Filters can add mood and contrast, creating thematic
styles for images.
Auto-Alignment and Cropping in Canva: Canva’s AI-based smart cropping
analyzes the main elements and focal points within an image, making cropping
adjustments that enhance composition. This helps maintain balance and visual appeal,
especially when resizing images for specific formats or layouts.
4. SAMPLE IMAGES AND RESULTS

Example images showing the applied techniques should be included here, demonstrating the
changes made by each tool. Each image could be processed with three different techniques
per tool to display the impact of each enhancement.

CODE SAMPLE FOR BACKEND (PYTHON/FLASK)

Here’s a simplified example of how the backend in Python Flask could look for the image
enhancer app.

>Python

>From Flask Import Flask, Request, Jsonify, Send _File

>From PIL Import Image

>Import Io

>Import Torch

>from model import Enhance Model # Import a pre-trained model (e.g., Super-
Resolution)

app = Flask(__name__)

>Model = Enhance model() # Load Your AI Model Here

>mode l.eval()@app.route('/upload', methods=['POST'])

>def upload _image():

>file = request .files['image']

>img = Image. open(file .stream)

>img. save("uploaded_image.jpg")

> return jsonify ({"message": "Image uploaded successfully"}), 200

>@app.route('/enhance', methods=['POST'])

>Def Enhance _Image():

 ># Load uploaded image

> img = Image .open("uploaded_image.jpg")

># Process image with the model

>input _tensor = preprocess (img)

>with torch.no _grad():

> enhanced _ img = model(input _tensor)

# Convert the result back to image

>Enhanced _Img = Postprocess(Enhanced _ Img)

>Buffer = Io .Bytesio()

> Enhanced _ Img. Save(Buffer, Format="JPEG")

> Buffer. Seek(0)

>Return Send _File(Buffer, Mime _type="Image/Jpeg")

Def Preprocess (Img):

# Convert image to a tensor or any format your model requires

return torch .from _numpy (np .array (img)).un squeeze(0).float()

Def Postprocess(Tensor):

# Convert Tensor Back To Image

> Img = Tensor .Squeeze(). Numpy()

> Return Image .Fromarray(Np.Uint8(Img))

if __name__ == '__main__':

> App .Run(Debug=True)

---------------------------------------------------------------------------------------------------------------

Frontend Sample (React)

Here’s how you might handle image upload and display the enhanced image using React.
>>java_ script

>>import React, { use State } from 'react';

>>function App() {

>>const [image, set Image] = use State(null);

>> const [enhanced Image, set Enhanced Image] = use State(null);

>>const handle Upload = async (event) => {

>> const file = event. target. files[0];

>>set Image (URL. create Object URL(file));

>> const form Data = new Form Data();

>> form Data. append("image", file);

>>await fetch("/upload", { method: "POST", body: form Data });

};

>>const handle Enhance = async () => {

>>const response = await fetch("/enhance", { method: "POST" });

>>const blob = await response. blob();

>> set Enhanced Image(URL.createObjectURL(blob));

};

>> return (

>> <div>

>> <input type="file" onChange={handleUpload} />

>> {image && <img src={image} alt="Original" />}

>> <button onClick={handleEnhance}>Enhance</button>

>> {enhancedImage && <img src={enhancedImage} alt="Enhanced" />}

>> </div>
 );

}>>export default App;

Final Notes

Testing: Run the app locally and test with a range of images to ensure functionality
and performance.
Optimize For Speed: Implement image compression and model optimizations to
improve processing time.
Error Handling: Add error messages for unsupported image types, failed uploads,
etc.

This outline and code samples provide a foundation for creating an AI photo enhancer app
accessible online.

5. CONCLUSION

AI image processing tools, such as Adobe Photoshop Express and Canva, offer a range of
accessible and intuitive features that improve image quality and visual appeal. Techniques
like auto-enhance, background removal, noise reduction, sharpness adjustment, and color
enhancement highlight how AI can streamline and elevate image editing tasks. These tools
empower users to produce professional-quality images without extensive technical skills,
marking a significant advancement in how digital media is created and consumed. AI-
powered image processing continues to evolve, with ongoing improvements in machine
learning algorithms, enabling even more precise and contextual aware enhancements. As
these tools develop, they will likely become even more integral to industries that rely on
high-quality visual content, from e-commerce to digital marketing and beyond.

6. REFERENCES

 Adobe Photoshop Express Documentation, Adobe Inc. Accessed 2024.

 Canva Help Center, Canva Inc. Accessed 2024.
 "A Comprehensive Guide to AI in Image Processing," Journal of Artificial
Intelligence, 2023.
 "Image Segmentation and Its Application in Background Removal," Computer Vision
Society, 2023.