This is MaskRCNN model trained for blurring advertising on the streets.

Getting Started

  • Using a model pre-trained on custom dataset to segment and blur objects in your own images or video.
# Detect and blur image or video

python3 (--image_path=/path/to/image.jpg | --video_path=/path/to/video.mp4) --model=/path/to/weigth.h5 
  • You can download dataset annotation to train your own model, the dataset consists of photos from Google street view. dataset


Python 3.4, TensorFlow 1.3, Keras 2.0.8 and other common packages listed in requirements.txt.


  1. Clone this repository
  2. Install dependencies
    pip3 install -r requirements.txt
  3. Run setup from the repository root directory
    python3 install
  4. Download pre-trained weights (mask_rcnn_ad.h5) from the releases page.