search

xggIoU / centernet_tensorflow_wilderface_voc

This is the unofficial implementation of the "CenterNet:Objects as Points".Just a simple try with self-modified shufflenetv2 and yolov3.If you want better results, you need more experiments.
MIT License
111 stars 25 forks source link

Keypoint Estimation #16

Open shreejalt opened 5 years ago

shreejalt commented 5 years ago

Hi @xggIoU, I am planning to use YOLOv3 as a backbone for feature extraction and want to do pose estimation instead of box detection. How can I leverage that use case in your training code? Also, what changes need to be done for offset loss as you are right now only predicting the class loss i.e heatmap loss and size loss. Can you help me out in solving this query?

Thanking You

xggIoU commented 5 years ago

Sorry, this depends on your own efforts. After all, there are still many changes that need to be made.