MedChaabane / DEFT

Joint detection and tracking model named DEFT, or ``Detection Embeddings for Tracking." Our approach relies on an appearance-based object matching network jointly-learned with an underlying object detection network. An LSTM is also added to capture motion constraints.
MIT License
269 stars 43 forks source link

Is there any script/code to run a DEFT inference directly on a mp4 video? #7

Closed jackhawa closed 3 years ago

MedChaabane commented 3 years ago

Thanks for your question. In the future, we will add code to run DEFT directly on a mp4 video. For now, If you want to test DEFT on your own dataset you can convert it into coco format and then you will be able to train and test DEFT on it. Thanks,