Crowdpose example

https://github.com/jin-s13/xtcocoapi/blob/master/demos/demo_crowdpose.py

import logging
import numpy as np
import faster_coco_eval
from faster_coco_eval import COCO, COCOeval_faster

print(f"{faster_coco_eval.__version__=}")

logging.root.setLevel("INFO")
logging.debug("Запись.")

faster_coco_eval.__version__='1.6.4'

gt_file = '../tests/dataset/example_crowdpose_val.json'
preds = '../tests/dataset/example_crowdpose_preds.json'

sigmas = np.array([
            .79, .79, .72, .72, .62, .62, 1.07, 1.07, .87, .87, .89, .89, .79,
            .79
        ]) / 10.0

cocoGt = COCO(gt_file)
cocoDt = cocoGt.loadRes(preds)
cocoEval = COCOeval_faster(cocoGt, cocoDt, 'keypoints_crowd', kpt_oks_sigmas=sigmas, use_area=False)
cocoEval.evaluate()
cocoEval.accumulate()
cocoEval.summarize()

cocoEval.stats_as_dict

INFO:faster_coco_eval.core.cocoeval:Evaluate annotation type *keypoints_crowd*
INFO:faster_coco_eval.core.cocoeval:COCOeval_opt.evaluate() finished...
INFO:faster_coco_eval.core.cocoeval:DONE (t=0.00s).
INFO:faster_coco_eval.core.cocoeval:Accumulating evaluation results...
INFO:faster_coco_eval.core.cocoeval:COCOeval_opt.accumulate() finished...
INFO:faster_coco_eval.core.cocoeval:DONE (t=0.00s).
INFO:faster_coco_eval.core.cocoeval: Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets= 20 ] = 0.788
INFO:faster_coco_eval.core.cocoeval: Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets= 20 ] = 0.988
INFO:faster_coco_eval.core.cocoeval: Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets= 20 ] = 0.731
INFO:faster_coco_eval.core.cocoeval: Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 20 ] = 0.822
INFO:faster_coco_eval.core.cocoeval: Average Recall     (AR) @[ IoU=0.50      | area=   all | maxDets= 20 ] = 1.000
INFO:faster_coco_eval.core.cocoeval: Average Recall     (AR) @[ IoU=0.75      | area=   all | maxDets= 20 ] = 0.778
INFO:faster_coco_eval.core.cocoeval: Average Precision  (AP) @[ IoU=0.50:0.95 | type=  easy | maxDets= 20 ] = 1.000
INFO:faster_coco_eval.core.cocoeval: Average Precision  (AP) @[ IoU=0.50:0.95 | type=medium | maxDets= 20 ] = 0.980
INFO:faster_coco_eval.core.cocoeval: Average Precision  (AP) @[ IoU=0.50:0.95 | type=  hard | maxDets= 20 ] = 0.412

{'AP_all': 0.7877215935879303,
 'AP_50': 0.9881188118811886,
 'AP_75': 0.7314356435643564,
 'AR_all': 0.8222222222222223,
 'AR_50': 1.0,
 'AR_75': 0.7777777777777778,
 'AP_easy': 1.0,
 'AP_medium': 0.9802,
 'AP_hard': 0.4116}

Orig Code

from xtcocotools.coco import COCO
from xtcocotools.cocoeval import COCOeval


cocoGt = COCO(gt_file)
cocoDt = cocoGt.loadRes(preds)
cocoEval = COCOeval(cocoGt, cocoDt, 'keypoints_crowd', sigmas, use_area=False)
cocoEval.evaluate()
cocoEval.accumulate()
cocoEval.summarize()

Orig result

loading annotations into memory...
Done (t=0.00s)
creating index...
index created!
Loading and preparing results...
DONE (t=0.00s)
creating index...
index created!
Running per image evaluation...
Evaluate annotation type *keypoints_crowd*
DONE (t=0.00s).
Accumulating evaluation results...
DONE (t=0.00s).
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets= 20 ] =  0.788
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets= 20 ] =  0.988
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets= 20 ] =  0.731
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 20 ] =  0.822
 Average Recall     (AR) @[ IoU=0.50      | area=   all | maxDets= 20 ] =  1.000
 Average Recall     (AR) @[ IoU=0.75      | area=   all | maxDets= 20 ] =  0.778
 Average Precision  (AP) @[ IoU=0.50:0.95 | type=  easy | maxDets= 20 ] = 1.000
 Average Precision  (AP) @[ IoU=0.50:0.95 | type=medium | maxDets= 20 ] = 0.980
 Average Precision  (AP) @[ IoU=0.50:0.95 | type=  hard | maxDets= 20 ] = 0.412