Yjaywang / hand-pose-detection-practice

hand-pose-detection-practice using MobileNet-SSD

Hand pose detection

This is my first object detection practice project, the final model will be converted to TensorFlow.js model and used by my video conferece website project (Meeting).

In this practice, some goals need to be achieved for my end application:

• Light-weighted
• Low computation consumption
• Low latency
• Acceptable accuracy

Based on these requirements, I choose MobileNet-SSD to train the model.

MobileNet Single Shot Detection (MobileNet-SSD)

Training data preparation

Image Class

• 9 classes

Labeling

• use LabelImg

Train/test ratio

• train: 90%; test: 10%

Training result

Reference

• https://www.researchgate.net/publication/343223517_A_Mobile-Based_Framework_for_Detecting_Objects_Using_SSD-MobileNet_in_Indoor_Environment
• https://www.analyticsvidhya.com/blog/2022/09/object-detection-using-yolo-and-mobilenet-ssd/
• https://www.researchgate.net/figure/MobileNet-SSD-AF-architecture-we-use-MobileNet-as-the-feature-extractor-network-and-SSD_fig7_324584455
• https://www.youtube.com/watch?v=pDXdlXlaCco
• https://www.youtube.com/watch?v=ZTSRZt04JkY
• https://www.tensorflow.org/tutorials/images/transfer_learning