Dashcam Anonymizer

This repository blurs human faces and license plates in images and videos, using a state-of-the-art object detection model, YOLOv8 by Ultralytics and is fine-tuned using images from the OpenImagesDatasetV7.

git clone this repo by

https://github.com/varungupta31/dashcam_anonymizer.git

Install the conda environment

#If on Linux
conda env create -f environment.yml
conda activate dashanon
#If on MacOS
conda env create -f envmacos.yml
conda activate dashanonmac

Blurring Images in a Directory

To blur all images in a directory,

Download the model from here and move it to model dir.

Update the configs/img_blur.yaml as required, and run the following command

python blur_images.py --config configs/img_blur.yaml

The resulting blur images will be stored in the directory specified in the YAML. Note: annot_txt folder will contain the YOLO detections in .txt format, converted to the VOC bounding-box format.

Blurring Videos in a Directory

Similar approach as above, now the command would be

python blur_videos.py --config configs/vid_blur.yaml

Note:

1. Make sure the opencv is installed via Conda. PIP installation, leaves out some libgc libraries, which causes issues in the videowriter codecs.
2. The configuration files are slightly different for videos and images. Make sure to choose and edit the correct ones depending upon the modality.
3. This is designed to process all the contents in a given directory at once. If the blurring is to be re-run, make sure to delete the runs directory, as it may lead to new file names within the runs, which will cause errors.
4. The blur_videos.py script currently expects the videos to be named numerically [1.mp4, 111.mp4]

If this repository helped you in a research project, Please cite the work

@software{dashcam_anonymizer,
  author = {Varun Gupta},
  month = {8},
  title = {{Dashcam Anonymizer}},
  url = {https://github.com/varungupta31/dashcam_anonymizer},
  version = {1.0.0},
  year = {2023}
}

and please consider,