Detecting and tracking person using deep_sort with yolov3. Resnet50 and pretrained weights are used for feature extraction. Another feature of my implementation includes face detection with MTCNN
I have no gpu PC.So I implement this in google colab.
To run project.ipynb, dowload all the folders and upload to your google drive.
In your google colab, folder destination should be like this
drive
- MyDrive
- [deep_sort_pytorch-master]
- [PyTorch-YOLOv3-master]
- [yolov3.weights]
- [feature_extraction.py]
- [project.ipynb]
Here is the demo :)
Code
Deep_sort_pytorch: https://github.com/ZQPei/deep_sort_pytorch YOLOv3: https://github.com/eriklindernoren/PyTorch-YOLOv3 Resnet50 from torchvision.models
Paper
Deep_sort_pytorch: https://arxiv.org/abs/1703.07402 YOLOv3: https://arxiv.org/abs/1804.02767 Resnet: https://arxiv.org/abs/1512.03385