Skip to content

Latest commit

 

History

History
24 lines (17 loc) · 676 Bytes

README.md

File metadata and controls

24 lines (17 loc) · 676 Bytes

MegaDetector Contained

Docker image for running the MegaDetector camera-trap object detection model.

Requirements

  • CUDA 11.0 (adjustable by changing base image)
  • docker
  • nvidia-docker (sudo apt install nvidia-docker2 on Ubuntu 20.04.1)

Building

docker build -t bencevans/megadetector .

Batch Detection

nvidia-docker run --rm -it -v "$(pwd)":/dataset bencevans/megadetector \
  python detection/run_tf_detector_batch.py \
  --recursive --output_relative_filenames --checkpoint_frequency 10000 \
  /workspace/blobs/4.1.0.pb /dataset/ /dataset/md.4.1.0.json