readme.md

Prepare the datasets

Download MS COCO 2017 dataset, LVIS V1 dataset and RefCOCO+ dataset and organize as follows:

--datasets
  |--coco
  |  |--annotations
  |  |--val2017
  |--lvis_v1
  |  |--annotations
  |  |--train2017
  |  |--val2017
  |--refcoco+
  |  |--image
  |  |--testA.tsv
  |  |--testB.tsv
  |  |--train.tsv
  |  |--val.tsv
  |--coco_groundingdino_correct_detection.json
  |--coco_groundingdino_misclassification.json
  |--coco_groundingdino_misdetect.json
  |--coco_florence-2_correct_detection.json
  |--coco_mask_rcnn_correct.json
  |--coco_yolo_v3_correct.json
  |--coco_fcos_correct.json
  |--coco_ssd_correct.json
  |--lvis_v1_rare_groundingdino_correct_detection.json
  |--lvis_v1_rare_groundingdino_misclassification.json
  |--lvis_v1_rare_groundingdino_misdetect.json
  |--refcoco_val_groundingdino_correct.json
  |--refcoco_val_groundingdino_mistake.json
  |--refcoco_val_florence-2_correct.json