Computer Vision: YOLO Custom Object Detection with Colab GPU

Packt via Coursera

Go to class Write review

Details

Go to class

Provider

Coursera
Pricing

Free Online Course (Audit)
Languages

English
Certificate

Paid Certificate Available
Duration & workload

7 hours 24 minutes
Sessions

On-Demand
Level

Beginner
Subtitles

English

Found in

Overview

Save Big on Coursera Plus. 7,000+ courses at $160 off. Limited Time Only!

Grab it

In this comprehensive course, you'll dive into the world of real-time object detection with YOLO, one of the most powerful algorithms for detecting objects in images and videos. The course begins with an introduction to YOLO and object detection, followed by setting up your development environment with Anaconda and installing essential libraries like OpenCV. A review of Python basics ensures you are equipped with the necessary programming knowledge before delving into convolutional neural networks (CNNs). Once your environment is ready, the course progresses into more advanced topics such as implementing YOLO for pre-trained object detection. You’ll explore practical examples, including detecting objects in images, videos, and live webcam feeds. The course then takes you through custom training with YOLOv4, where you will learn to collect and label data, train-test split, and prepare Darknet for training your own models. Each phase of custom training is covered step by step, including synchronization with Google Colab and Drive, testing Darknet, and fine-tuning the training process. By the end of the course, you'll be adept at training YOLO models for specific use cases, including the detection of various objects and even custom challenges such as COVID-19 detection. Along the way, you'll troubleshoot common issues like GPU usage limits in Colab and explore real-world case studies to solidify your understanding. No prior knowledge of YOLO is required, but a basic understanding of machine learning concepts will be helpful. This course is designed for data scientists, machine learning engineers, and computer vision enthusiasts who are familiar with Python programming.

Syllabus

Course Introduction and Table of Contents

In this module, we will introduce the course content and outline the key concepts you'll be learning. This section will provide an overview, helping you understand the course structure and what to expect as you progress.

Introduction to You Only Look Once (YOLO) Object Detection

In this module, we will dive into the basics of YOLO, a state-of-the-art object detection algorithm. You'll learn about its scope, importance, and why it's widely used in various computer vision applications.

Environment Setup - Installing Anaconda

In this module, we will guide you through installing and setting up Anaconda, a popular platform for managing Python environments. You'll learn how to prepare your system for running the course projects.

Python Basics

In this module, we will cover fundamental Python programming concepts, including flow control, data structures, and functions. These basics are crucial for developing and understanding the custom YOLO model later in the course.

Installing OpenCV Library

In this module, we will walk you through the installation of the OpenCV library, a key tool for image processing and computer vision. You'll ensure your environment is ready for the practical tasks ahead.

Introduction to Convolutional Neural Networks (CNNs)

In this module, we will introduce Convolutional Neural Networks (CNNs), the backbone of many modern computer vision applications. You'll gain insights into how CNNs function and their relevance to YOLO.

You Only Look Once (YOLO) Pre-Trained Object Detection from Image

In this module, we will guide you through using a pre-trained YOLO model to detect objects in images. You'll learn how to perform this task step-by-step, gaining hands-on experience with the YOLO algorithm.

You Only Look Once (YOLO) Pre-Trained Object Detection from an Image – Non-Maximum Suppression (NMS)

In this module, we will explore Non-Maximum Suppression (NMS), a technique used to improve object detection accuracy in YOLO. You'll see how NMS helps eliminate redundant detections, refining the final output.

You Only Look Once (YOLO) Pre-Trained Object Detection from a Real-Time Webcam Video

In this module, we will demonstrate how to perform real-time object detection using a webcam and a pre-trained YOLO model. You'll learn to adapt YOLO for live video feeds, enhancing your practical skills.

You Only Look Once (YOLO) Pre-Trained Object Detection from a Pre-Saved Video

In this module, we will show you how to apply YOLO to detect objects in pre-saved video files. You'll explore the nuances of video-based detection and how to optimize the model for such tasks.

Introduction to the Custom-Trained You Only Look Once (YOLO) Model

In this module, we will introduce you to the process of custom training a YOLO model. You'll learn about the advantages of customizing YOLO for specific tasks and get an overview of the training process.

YOLOv4 Custom Training Phase 1 – Preparing Darknet

In this module, we will focus on setting up the Darknet environment, a key step in custom training YOLOv4 models. You'll download the necessary weights and prepare your system for the training process.

You Only Look Once v4 (YOLOv4) Custom Training Phase 2 - Data Collection

In this module, we will guide you through the data collection process for training a YOLOv4 model. You'll learn how to gather and organize data effectively, ensuring your training dataset is robust.

You Only Look Once v4 (YOLOv4) Custom Training Phase 2 - Image Labelling

In this module, we will cover the image labeling process, a critical step in preparing your dataset for YOLOv4 training. You'll use labeling tools to create accurate and consistent annotations for your images.

You Only Look Once v4 (YOLOv4) Custom Training Phase 2 - Train Test Split

In this module, we will explain the concept of train-test splitting, essential for evaluating the performance of your YOLOv4 model. You'll learn how to balance your data to achieve optimal training results.

You Only Look Once v4 (YOLOv4) Custom Training Phase 2 - Data Preparation

In this module, we will focus on the final stages of preparing your dataset for YOLOv4 training. You'll apply preprocessing techniques to ensure your data is ready for the training phase.

You Only Look Once v4 (YOLOv4) Custom Training Phase 3 – Data Sync

In this module, we will demonstrate how to sync your data with Google Drive and connect it to Colab. You'll learn how to manage your files efficiently, ensuring smooth operation during model training.

You Only Look Once v4 (YOLOv4) Custom Training Phase 4 - Compile and Test Darknet

In this module, we will guide you through compiling and testing Darknet, the framework used for YOLOv4 training. You'll learn to resolve any issues that may arise during the setup process.

You Only Look Once v4 (YOLOv4) Custom Training Phase 5 - Chart and Training Progress Analysis

In this module, we will explore how to monitor and analyze the training progress of your YOLOv4 model. You'll use charts and metrics to assess performance and make necessary adjustments.

You Only Look Once v4 (YOLOv4) Custom Training Phase 5 - Finalizing Training Download Weights

In this module, we will cover the final steps of YOLOv4 training, including downloading and saving the model weights. You'll learn how to complete the training process and prepare your model for deployment.

Colab GPU Usage Limit Issue

In this module, we will discuss the GPU usage limits in Google Colab and how they may affect your YOLOv4 training. You'll learn strategies to manage these limits and keep your training process uninterrupted.

OpenCV Upgrade for You Only Look Once v4 (YOLOv4)

In this module, we will guide you through upgrading OpenCV to ensure compatibility with YOLOv4. You'll learn how to perform the upgrade and resolve any issues that may arise.

You Only Look Once v4 (YOLOv4) Pre-Trained Object Recognition from an Image and a Video

In this module, we will demonstrate how to use a pre-trained YOLOv4 model to detect objects in both images and videos. You'll explore the model's versatility and practical uses in various scenarios.

You Only Look Once v4 (YOLOv4) Custom Coronavirus Detection from an Image

In this module, we will show you how to train a YOLOv4 model to detect coronavirus in images. You'll learn the nuances of customizing YOLOv4 for specialized detection tasks.

You Only Look Once v4 (YOLOv4) Custom Coronavirus Detection from a Video

In this module, we will focus on applying a custom-trained YOLOv4 model to detect coronavirus in videos. You'll gain experience in adapting image-based models for video analysis.

Other Sample Real-World Case Studies

In this module, we will present additional real-world case studies demonstrating the application of YOLO in different industries. You'll see how the concepts learned can be applied to solve real-world challenges.