event_tracking

Visual tracking with event cameras by using E2VID and YOLO networks.

User requirements

Our code has been tested on Ubuntu 16.04, in an Anaconda3 environment with Python 3.8 and the following configuration:

conda install pytorch==1.7.1 torchvision==0.8.2 torchaudio==0.7.2 cudatoolkit=11.0 -c pytorch
conda install pandas
conda install scipy
pip install opencv-python

To run the event tracker, the E2VID repository needs to be cloned, along with their pretrained model. This can be found here.

A pretrained model of the YOLOv5 network is used by loading from the Pytorch Hub. See details here.

Run the tracker

python estimation_event_tracker.py \
--input e2vid/rpg_e2vid/data/events.txt \
--output_folder output/ \
--det_class 0  \
--num_events_per_pixel 0.5

References

This work uses the E2VID and YOLOv5 neural networks.

• H. Rebecq, R. Ranftl, V. Koltun, and D. Scaramuzza, “High speed andhigh dynamic range video with an event camera,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 43, no. 6, pp. 1964–1980, 2021.
• G. Jocher et al., “YOLOv5n ’Nano’ models, Roboflow integration, Tensor-Flow export, OpenCV DNN support,” doi:10.5281/zenodo.5563715,2021.