A series of notebooks to dive deep into popular datasets for object detection and learn how to train Detectron2 on a custom dataset.
- Notebook 00: Install Detectron2
- Notebook 01a: Load and read COCO dataset with
COCO PythonAPI
andGluonCV
- Notebook 01b: Load and read Open Images v5
- Notebook 02: Format Open Images annotations for Detectron2
- Notebook 03: Train Detectron2 on a small dataset of Camera and Tripod
- Notebook 04: Put all steps together: download Open Images v6 and train Detectron2 on a large dataset of 11 musical instruments.
- Notebook 05: Apply Detectron2 to solve a real world challenge (Kaggle Global Wheat Detection Competition)
From Notebook 03: Camera and Tripod