Beginning of dialog window. Escape will cancel and close the window.
End of dialog window.
详情
字幕
An Applied Case for Fast Object Detection Embedded on NVIDIA Jetson Devices with TensorRT Optimization
, Machine Learning Engineer, Preste
High accuracy and performance are the two main components of any successful AI product. Engineers create and test models to reach a final goal by balancing these two factors. We'll present a real case of an AI project for fast object detection and analytics embedded on NVIDIA Jetson devices. Learn how to reach 120 FPS live video processing with limited resources, and the NVIDIA TensorRT accelerator's role in it. We'll share insights on how we managed the corner cases with a tailored dataset and tackled concrete implementation issues. We'll also discuss the result, benefits, and advantages of moving from a sensor-based to a deep learning-based solution. You'll understand how to design, train, and optimize a neural network depending on concrete tasks.