Hand Detection in Tensorflow
A hand object detection using ObjDet. The object detector in the example uses SSD300 on top of MobileNetV2 applying transfer learning from a pretrained model on MS COCO. The model is trained with the Egohands Dataset. The main goal is to demonstrate the easy use of objdet and to build a fast detector ready for realtime applications. A later application for tracking can be found in ObjTracking (Coming soon).
Feel free to help me improve this project!
Installation
python -m venv env
source env/bin/activate
pip install -r requirements.txtDownload the Egohands Dataset in the data folder and execute the following:
unzip egohands_data.zip -d egohands && rm egohands_data.zipDownload the CMU Hand DB in the data folder and execute the following:
unzip hand_labels.zip && mv hand_labels handdb && rm hand_labels.zipGoogle Colab
In order to run the notebook in Google Colab