GOOGLE AI-ML VIRTUAL INTERNSHIP

PRESENTED BY:
P.Hanish
Internal Gudie:
(213J1A4438)
MR. P.Ravi(Ph.D)
Duration:2 Months ASSISTANT PROFESSOR
Provided By: EDU SKILLS
AICTE EDUSKILLS
GOOGLE AI-ML
VIRTUAL
INTERNSHIP
CONTENTS
1.Introduction
2.Program neural networks with TensorFlow
3.Object detection
4.Product image search
5.Image classification
6.Conclusion
 INTRODUCTION
Today's world needs understanding and handling of visual information in
almost every field, for instance: self-driving technology, e-health, e-commerce
among others. Computer vision, a branch of artificial intelligence, involves the
capacity of computers to process visual information and has a number of applications
that improve user and business processes.
In this project we will focus on the application of TensorFlow in solving
important computer vision problems namely: object detection, image classification
and product search. TensorFlow is a broad open- source library created by Google
that allows the easy integration of advanced solutions into existing architectures.
Self-driving cars and the pattern recognition technologies require object
detection whereas image classification is useful in enhancing medical images as well
as content security. Besides that, visual product search has also gained importance
especially in online shopping where images are used to source for products instead of
written text.
Program
neural
networks with
TensorFlow
Computer Vision and Convolutional Neural Networks (CNNs)Computer Vision is a
field that enables machines to interpret and understand visual data, such as images and
videos. It allows computers to perform tasks similar to human vision, such as:
 Recognizing
 Objects Detecting
 Motion Understanding scenes.
This technology has driven advancements in various fields, including:
 Facial recognition
 Self-driving cars
 Medical imaging
Convolutions in Image Processing. A convolution is a key mathematical operation used
to extract patterns from images. It involves applying filters (kernels) across the image to
detect features like edges, textures, and corners. These filters slide over the image,
capturing essential details that contribute to feature extraction.
 Convolutional Neural Networks (CNNs)CNNs are specialized deep learning models that
excel in processing visual data. They are structured with the following layers:
 Convolutional Layers – Extract key features by applying filters to the input image.
 Pooling Layers – Reduce data dimensionality, retaining only the most significant information.
 Fully Connected Layers – Map the extracted features to the output, such as image classification
or object detection.
CNNs are powerful because they capture hierarchical patterns in images—from simple elements
like edges to more complex shapes and objects. Handling Large Datasets with CNNs Working with
large datasets presents challenges, particularly in terms of computational resources. However, two
effective techniques address this:
 Data Augmentation: Expands the dataset by modifying images (e.g., rotating, flipping, or
scaling).Helps prevent overfitting without requiring additional data collection.
 Transfer Learning: Uses pre-trained CNN models and fine-tunes them for specific tasks. Saves
time and computational power by leveraging existing knowledge from earlier models. These
techniques enable CNNs to achieve high accuracy in complex tasks like image classification
and object detection, even with large datasets.
Object
Detection
Key Concepts and Applications: Object Detection goes beyond image classification
by identifying multiple objects and drawing bounding boxes around them. It is
crucial for applications like:
 Facial Recognition
 Autonomous Driving
 Augmented Reality (AR)
Tools for Object Detection: ML Kit and TensorFlow Lite are commonly used tools
for developing mobile object detection applications. Features include:
 Real-time detection (even offline) using the Object Detection and Tracking
(ODT) API.
 Custom model training with TensorFlow Lite Model Maker for unique object
detection tasks.
Applicable in: Mobile apps, autonomous systems, and product search solutions.
Object Detection Process Overview:
User Interface Setup: Users can open the camera, take a photo, or use preset
images for testing. Images are converted into Bitmap format for processing.
Object Detection Setup: Create a detector instance and configure options
(e.g., single/multi-object detection, classification).
Processing and Results: Image is processed asynchronously, with results
(bounding boxes and labels) sent back via a callback function. Enables
efficient real-time response for applications.
Product
image search
Definition:
Detects objects in images and searches for visually similar products online.
How it Works:
Uses ML Kit's Object Detection and Tracking API to detect and track objects. Sends
the selected object to a backend service like Google’s Vision API Product Search to
compare with a product catalog.
Output:
Returns matching or visually similar products.
Features and Applications Model Integration:
Pre-trained models or custom models via TensorFlow Lite (e.g., fashion, home
goods).
Use Case Examples:
If an image shows a table and chairs, users can select specific objects to search
individually.
Google Lens:
Real-time product search using smartphone cameras.
Benefits for E-commerce:
 Enhanced Shopping Experience: Users find products by taking pictures
instead of typing descriptions. Simplifies product discovery and improves
purchase convenience.
 Retailer Integration: Retailers upload product images, enabling catalog
matching and visual search. Supports personalized product
recommendations based on user inputs.
IMAGE
CLASSIFICATION
Definition:
A computer vision task that categorizes images into predefined labels or classes based
on their content.
Powered by:
Machine learning algorithms using deep learning models like Convolutional Neural
Networks (CNNs).
Goal: Assign a single label to an image (e.g., identifying a dog, cat, or car).How Image
Classification Works.
Data Preparation: Collect a labeled dataset where images are organized into folders
(e.g., "cat," "dog," "car").
Model Training: Images pass through multiple CNN layers.
Convolutional layers extract features (e.g., edges, textures).Pooling layers reduce
dimensionality for efficient processing. Fully connected layers generate class
probabilities.
Feature Extraction: The model learns to recognize patterns (e.g., fur patterns or
snout shapes) to distinguish between objects.
Classification: Given a new image, the model analyzes it and assigns a label with a
probability score.
Evaluation: The model’s performance is assessed using metrics like:
Accuracy
Precision
Recall
 CONCLUSION

Finally, I completed a capstone project focused on using Google Cloud AI and ML

tools, where I implemented a machine learning solution to a real business problem. This project
allowed me to apply everything I learned, from data preprocessing and model building to
deployment. This practical experience has prepared me for real-world challenges where I can
apply AI and ML to drive meaningful results.
Overall, this virtual internship has not only expanded my technical knowledge but also
solidified my passion for pursuing a career in artificial intelligence and machine learning. The
combination of theoretical learning and practical application through these courses has
significantly enriched understanding of AI and ML. I am excited to leverage this knowledge as
I continue to explore the field and contribute to innovative solutions in the AI/ML domain.
Thank you