This is a project for UWindsor's 4740
This project is a Graph based convolutional using super pixels to attempt varing tasks like classification/segmentation/detection/etc...
You need to install Python of at least 3.11
Personally, I just use the extension on VScode, and that is what I reccommend
venv comes with Python, but you'll want to remember to Selec the venv as the Notebooks Kernel and to run any terminal commands from within the environment
As you may guess I cant just shove an entire data set into a repo, you'll need:
- The COCO '2017 Train Images'
- Place them into
Data/train2017
- Place them into
- The COCO '2017 Val Images'
- Place them into
Data/val2017
- Place them into
- The COCO '2017 Test Images'
- Place them into
Data/test2017
- Place them into
- A 'Objects_train.csv' file of parsed annotations, which you can download with the link
- Place it into ParsedAnnotations
- A 'Segmentations_train.json' file of parsed annotations, which you can download with the link
- Place it into ParsedAnnotations