Process This: Efficient human pose estimation using the YOLO-Pose model and TI processors
In this episode of Process This, learn how to easily achieve efficient Human Pose Estimation in an embedded application using our free patent-pending Human Pose Estimation model. Human Pose Estimation is a popular computer vision task used in many products, such as surveillance, medical treatment, robotics, sports therapy, and exercise posture analysis. Our innovative YOLO-pose model reduces latency by 2.5 times. No embedded edge AI expertise or any AI tools are needed to use our model. We'll show you how to use our free cloud tools to get started.
Webinar topics include:
- Introduction to Human pose estimation.
- Target Applications: Video surveillance, fall detection, Robotics.
- Human pose estimation offering in TI Model Zoo.
- Yolo-pose model architecture overview based on YOLOv5 and YOLOX.
- Optimizations for TI’s deep learning accelerator.
- Compiling the model using open-source run time.
- Model performance/accuracy benchmarking.
- Evaluation on Free Cloud tool.