search

qqlu / Entity

EntitySeg Toolbox: Towards Open-World and High-Quality Image Segmentation
Other
692 stars 58 forks source link

evaluation results not good #2

Closed gezhaoDL closed 3 years ago

gezhaoDL commented 3 years ago

results of pretrained model mit_b5_1x.pth on COCO val set are not good.Here is my command:

python projects/EntitySeg/train_net.py --config-file projects/EntitySeg/configs/entity_mit_b5_1x.yaml --num-gpus 1 --eval-only MODEL.WEIGHTS data/models/mit_b5_1x.pth MODEL.CONDINST.MASK_BRANCH.USE_MASK_RESCORE "True"

and the results:

[08/11 16:11:11 d2.engine.defaults]: Evaluation results for coco_2017_val_entity in csv format: [08/11 16:11:11 d2.evaluation.testing]: copypaste: Task: bbox [08/11 16:11:11 d2.evaluation.testing]: copypaste: AP,AP50,AP75,APs,APm,APl [08/11 16:11:11 d2.evaluation.testing]: copypaste: 0.0010,0.0019,0.0008,0.0004,0.0014,0.0013 [08/11 16:11:11 d2.evaluation.testing]: copypaste: Task: segm [08/11 16:11:11 d2.evaluation.testing]: copypaste: AP,AP50,AP75,APs,APm,APl [08/11 16:11:11 d2.evaluation.testing]: copypaste: 0.0006,0.0015,0.0005,0.0002,0.0008,0.0009

gezhaoDL commented 3 years ago

i change the line self._coco_api = COCO(json_file, cfg.TEST.CLASS_AGNOSTIC) to self._coco_api = COCO(json_file) in the file _path/to/EntitySeg/entityseg/evaluator/entityevaluation.py

qqlu commented 3 years ago

You should use our cocoapi because the default cocoapi will consider the category. We kept the category information in the JSON file of our ground truth.

gezhaoDL commented 3 years ago

You should use our cocoapi because the default cocoapi will consider the category. We kept the category information in the JSON file of our ground truth.

Thanks for reply.I used modified_cocoapi and got correct results.

[08/11 17:24:03 d2.engine.defaults]: Evaluation results for coco_2017_val_entity in csv format: [08/11 17:24:03 d2.evaluation.testing]: copypaste: Task: bbox [08/11 17:24:03 d2.evaluation.testing]: copypaste: AP,AP50,AP75,APs,APm,APl [08/11 17:24:03 d2.evaluation.testing]: copypaste: 44.2095,64.7040,46.4046,17.7186,39.8511,58.1466 [08/11 17:24:03 d2.evaluation.testing]: copypaste: Task: segm [08/11 17:24:03 d2.evaluation.testing]: copypaste: AP,AP50,AP75,APs,APm,APl [08/11 17:24:03 d2.evaluation.testing]: copypaste: 38.7154,61.7845,40.6139,14.8040,34.7366,50.7424