# category_id = result["category_id"]
# # assert category_id < num_classes, (
# # f"A prediction has class={category_id}, "
# # f"but the dataset only has {num_classes} classes and "
# # f"predicted class id should be in [0, {num_classes - 1}]."
# # )
# result["category_id"] = reverse_id_mapping[category_id]
# seg_results.append(result)
#
# if not os.path.exists('{}/results'.format(args.result_dir)):
# os.mkdir('{}/results'.format(args.result_dir))
#
# with open('{}/results/{}_seg_results.json'.format(args.result_dir, args.data_type), 'w') as f_det:
# json.dump(seg_results, f_det)
if 'test' not in args.data_type:
if not os.path.exists('{}/results/plot'.format(args.result_dir)):
os.mkdir('{}/results/plot'.format(args.result_dir))
print(
'---------------------------------------------------------------------------------'
)
print('Running COCO segmentation val17 evaluation ...')
coco_pred = coco.loadRes('{}/results/{}_seg_results.json'.format(
args.result_dir, args.data_type))
coco_eval = COCOeval(coco, coco_pred, 'segm')
coco_eval.evaluate()
coco_eval.accumulate()
coco_eval.summarize()
coco_eval.analyze(
save_to_dir='{}/results/plot'.format(args.result_dir))
