Object Tracking Project

Uses a FasterRCNN with ResNet 50 backbone for object detection, and DeepSORT for object tracking.

COCO trained weights as well as COCO trained Fudan pedestrian dataset fine-tuned weights can be used.

Usage:

Create conda environment and activate:

    conda env create -f environment.yml
    conda activate objtracking

Explicitly define model and device:

    python main.py -i "./video/input.mp4" -o "./output/output.mp4" --model coco --display --verbose --device cuda

Model: coco, Device: automatic

    python main.py -i "./video/input.mp4" -o "./output/output.mp4" --display --verbose

Model: coco finetuned with Fudan pedestrian dataset(tracks only pedestrians), Device: automatic

    python main.py -i "./video/input.mp4" -o "./output/output.mp4" --model fudan