search

ilovepose / DarkPose

Distribution-Aware Coordinate Representation for Human Pose Estimation
https://ilovepose.github.io/coco
Apache License 2.0
550 stars 80 forks source link

How to test the model on COCO test-dev set 2017 #15

Closed Ozone-oo closed 3 years ago

Ozone-oo commented 3 years ago

I want to test model on test-dev set, but the code tool/test.py is only for testing on validation set (5000 images):

So, I have a question: How to test the model on COCO test-dev set 2017 to submit the json file to codalab server? I'm looking forward from you. Thank you for your great work!

hbin-ac commented 3 years ago

1、Download the test2017 dataset and image info. 2、Download the detection result on test-dev2017. 3、Modify the yaml file as follow DATASET: TEST_SET: 'test-dev2017' TEST: COCO_BBOX_FILE: 'data/coco/person_detection_results/COCO_test-dev2017_detections_AP_H_609_person.json' USE_GT_BBOX: false

Ozone-oo commented 3 years ago

1、Download the test2017 dataset and image info. 2、Download the detection result on test-dev2017. 3、Modify the yaml file as follow DATASET: TEST_SET: 'test-dev2017' TEST: COCO_BBOX_FILE: 'data/coco/person_detection_results/COCO_test-dev2017_detections_AP_H_609_person.json' USE_GT_BBOX: false

Thank you！！！