<a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/us

Strange behavior of weighted_boxes_fusion about weighted-boxes-fusion HOT 5 CLOSED

zfturbo commented on July 4, 2024

Strange behavior of weighted_boxes_fusion

from weighted-boxes-fusion.

Comments (5)

ZFTurbo commented on July 4, 2024 1

x1 must be lower than x2 and y1 must be lower than y2. I will fix it next release.

from weighted-boxes-fusion.

ZFTurbo commented on July 4, 2024

Coordinates for boxes expected to be normalized e.g in range [0; 1]. Order: x1, y1, x2, y2.

I will add warning on such cases.

from weighted-boxes-fusion.

shonenkov commented on July 4, 2024

@ZFTurbo no, it doesnt help!

wbf_boxes, wbf_scores, wbf_labels = ensemble_boxes.ensemble_boxes_wbf.weighted_boxes_fusion(
    (np.array(boxes)/511).tolist(),
    scores, 
    labels, 
    weights=None, 
    iou_thr=0.43, 
    skip_box_thr=0.44
)
wbf_boxes.shape
>>> (16, 4)

from weighted-boxes-fusion.

shonenkov commented on July 4, 2024

@ZFTurbo thank you! I will try it. I am looking forward to the new release :)

from weighted-boxes-fusion.

shonenkov commented on July 4, 2024

@ZFTurbo thank you! it works for me, I have used temporary "hardcode" (maybe it helps anyone):

   def prepare_boxes(boxes):
        result_boxes = boxes.copy()
        result_boxes[:,0] = np.min(boxes[:, [0,2]], axis=1)
        result_boxes[:,2] = np.max(boxes[:, [0,2]], axis=1)
        result_boxes[:,1] = np.min(boxes[:, [1,3]], axis=1)
        result_boxes[:,3] = np.max(boxes[:, [1,3]], axis=1)
        return result_boxes