Learn how to train and implement custom object detection using a Jetson Nano in this comprehensive video tutorial. Discover the process of capturing robot images, labeling them, and training a Single Shot Detection Network (MobileNet SSD) to detect robots in a live video stream. Explore computer vision models, object detection techniques, and the model preparation process, including working with Pascal VOC format and creating label files. Gain insights into machine learning concepts, image processing, and neural networks. Follow along with a step-by-step demonstration, learn essential commands, and understand the XML format for annotations. The tutorial concludes with a Q&A session, providing additional insights and clarifications on custom object detection with Jetson Nano.
Overview
Syllabus
Intro
Session Overview
What we're shooting for
Computer Vision Models
What is Object Detection
Model Preparation Process
Pascal VOC
Create a labels file
Capture Assets
Training and Testing
How does Machine Learning Work
Image Processing
Neural Networks
How does Machine Learning actually work though?
Single-Shot Detection Network
Train the model
Things I found out the hard way
Followed the Hello AI World from Nvidia website
Demo
Commands used
Annotations file - XML Format
Training the model
00 Q&A Session
Taught by
Kevin McAleer
Related Courses
-
Jetson AI Fundamentals - Training Object Detection Models
-
Computer Vision: YOLO Custom Object Detection with Colab GPU
-
Real-Time Object Detection in Python on Jetson Nano
-
AI on the Jetson Nano - Installing NVIDIA Object Detection and Inference Tools
-
AI on the Jetson Nano - Face and Eye Detection with Haar Cascades in OpenCV
-
Jetson AI Fundamentals - Training Image Classification Models
