This is a [BVLC Caffe](https://github.com/BVLC/caffe) fork that is intended for deployment of action detection and action recognition models for smart classroom use-case. You can define own list of possible actions (see annotation file [format]() and steps for model training to change the list of actions) but this repository shows example for 3 action classes: standing, sitting and raising hand.
This is a [BVLC Caffe](https://github.com/BVLC/caffe) fork that is intended for deployment of action detection and action recognition models for smart classroom use-case. You can define own list of possible actions (see annotation file [format]() and steps for model training to change the list of actions) but this repository shows example for 3 action classes: standing, sitting and raising hand.
...
@@ -135,8 +136,8 @@ On next stage we should train the Action Recognition (AR) model which reuses det
...
@@ -135,8 +136,8 @@ On next stage we should train the Action Recognition (AR) model which reuses det
where `pd_weights_path.cafemodel` - weights of trained PD model (see [previous](#person-detection-training) section) and `ar_init_weights_path.cafemodel` - output path to init weights for AR model.
where `pd_weights_path.cafemodel` - weights of trained PD model (see [previous](#person-detection-training) section) and `ar_init_weights_path.cafemodel` - output path to init weights for AR model.
where `action_detection_snapshot_path.caffemodel` is a selected snapshot of AR training procedure and `action_detection_inference_path.caffemodel` - path to save final model parameters.
where `action_detection_snapshot_path.caffemodel` is a selected snapshot of AR training procedure and `action_detection_inference_path.caffemodel` - path to save final model parameters.