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.
PythonMIT
Issues
- 0
Motion prediction model
#31 opened by yhifny - 4
- 0
packege issue
#30 opened by layeqaliali - 1
- 1
- 1
- 0
About the trajectory_dataset.
#27 opened by xiaocc612 - 3
- 0
Inference Speed
#25 opened by JaiDoshi - 0
Full metrics on nuScenes validation split
#24 opened by a1600012888 - 0
How can I use the model to predict my own video
#22 opened by Linkcy97 - 2
- 0
- 1
balancing weights for loss
#19 opened by Chenzhaowei13 - 0
- 0
Occlusion tracking evaluation
#18 opened by cdiazruiz - 2
- 1
- 1
Occlusion Handling and Testing Phase.
#11 opened by SamihaSara - 1
No module named 'opts'
#13 opened by jka-zed - 0
Training and evaluating step failed
#12 opened by jka-zed - 4
- 2
Changing the backbone
#5 opened by HilmiiKumdakci - 8
Unable to Run ./mot17_tracking.sh
#6 opened by ryanmaxwell96 - 1
- 1
How to train on COCO/own dataset?
#8 opened by Hiddenfire21 - 1
- 5
RuntimeError: Error compiling objects for extension from ./make.sh for DCNv2
#4 opened by ryanmaxwell96 - 1
Inference
#2 opened by AlexeySrus - 1
- 2