def eval(num_imgs):
# annotation_path = '/home/yjr/DataSet/COCO/2017/test_annotations/image_info_test2017.json'
annotation_path = '/home/yjr/DataSet/COCO/2017/test_annotations/image_info_test-dev2017.json'
# annotation_path = '/home/yjr/DataSet/COCO/2017/annotations/instances_train2017.json'
print("load coco .... it will cost about 17s..")
coco = COCO(annotation_path)
imgId_list = coco.getImgIds()
if num_imgs != np.inf:
imgId_list = imgId_list[:num_imgs]
faster_rcnn = build_whole_network_batch.DetectionNetwork(
base_network_name=cfgs.NET_NAME, is_training=False)
save_dir = os.path.join(cfgs.EVALUATE_DIR, cfgs.VERSION)
eval_with_plac(det_net=faster_rcnn,
coco=coco,
imgId_list=imgId_list,
out_json_root=save_dir,
draw_imgs=True)
print("each img over**************")
final_detections = []
with open(os.path.join(save_dir, 'coco2017test_results.json'), 'w') as wf:
for imgid in imgId_list:
f = open(os.path.join(save_dir, 'each_img', str(imgid) + '.json'))
tmp_list = json.load(f)
# print (type(tmp_list))
final_detections.extend(tmp_list)
del tmp_list
f.close()
json.dump(final_detections, wf)
def cocoval(detected_json, eval_json):
eval_gt = COCO(eval_json)
eval_dt = eval_gt.loadRes(detected_json)
cocoEval = COCOeval(eval_gt, eval_dt, iouType='bbox')
# cocoEval.params.imgIds = eval_gt.getImgIds()
cocoEval.evaluate()
cocoEval.accumulate()
cocoEval.summarize()
