fredzzhang / vcoco

Utilities for the dataset V-COCO in Python3

V-COCO

Utilities for the human-object interaction detection dataset V-COCO

Supported Utilities

• A command-line style dataset navigator
• Large-scale visualisation in web page
• Generate object detections with Faster R-CNN
• Visualise detected objects
• Evaluate object detections

Installation Instructions

1. Download the repo with git clone https://github.com/fredzzhang/vcoco.git
2. Download the train2014 and val2014 partitions of the COCO dataset
   1. If you have not downloaded the dataset before, run the following script

   cd /path/to/vcoco
   bash download.sh

   2. If you have previsouly downloaded the dataset, simply create a soft link. Note that

   cd /path/to/vcoco
   ln -s /path/to/coco ./mscoco2014

3. Install the lightweight deep learning library Pocket
4. Make sure the environment you created for Pocket is activated. You are good to go!

Miscellaneous

• The implementation of the dataset class can be found in vcoco.py
• The script that generated the annotation files can be found in utilities/generate_annotations.py

License

MIT License