Build a Custom Object Detector in Live Video using TensorFlow
- Artificial Intelligence
- Computer Vision
- Project length: 3h 12m
Ever wanted to make something that could detect an object of your choice? And if I told you that it could detect that even in live videos. Let us bring that imagination to reality with this project.
This tutorial will cover all the details (resources, tools, languages etc) that are necessary to build a complete and operational custom object detector for a live video* You will be guided through all the steps and concepts, starting from the basic ones like setting up the right tools and frameworks to the more advanced topics related to the development. And ultimately you will be able to detect your own objects without any difficulty.
What are the requirements?
- Python basics
- Basic neural network concepts
- And the most important is: a desire to learn
What is the target audience?
- Learners who want to enhance their knowledge in computer vision.
- Entrepreneurs who want to employ object detection in their tasks.
- This course will help the students who are doing their final projects
Session 1: Installation
- Session 1.1 Downloading required dependencies
- Session 1.2 Compiling
- Session 1.3 Jupyter notebook
- Session 1.4 Running our first object detection
Session 2: Detection in live Video
- Detection in live video
Session 3: Data Preparation
- Session 3.1 Data Augmentation Introduction
- Session 3.2 Data Augmentation
- Session 3.4 Bounding boxes setup
- Session 3.5 Bounding boxes
Session 4: Create tf Records
- Session 4.1: Convert data into array
- Session 4.2.1: Convert data into csv files
- Session 4.2.2: Convert data into csv files 2
- Session 4.3.1: Create tf records 1
- Session 4.3.2: Create tf records 2
Session 5: Custom Object Detection in Live Video
- Session 5.1: Setting up config files
- Session 5.2.1: Training
- Session 5.2.2: Training 2
- Session 5.3: How to view tensorboard
- Session 5.4: Custom Object Detection in Live Video