ValueError: operands could not be broadcast together with shapes (0,) (1,3)

[sailohitaksh-cryptic]

When testing with a certain video the code is giving this error.

Traceback (most recent call last):
File "D:\Documents\Ripik\truck_detect\main.py", line 31, in
track_ids = mot_tracker.update(np.asarray(detections_))
File "D:\Documents\Ripik\truck_detect\sort\sort.py", line 232, in update
matched, unmatched_dets, unmatched_trks = associate_detections_to_trackers(dets,trks, self.iou_threshold)
File "D:\Documents\Ripik\truck_detect\sort\sort.py", line 163, in associate_detections_to_trackers
iou_matrix = iou_batch(detections, trackers)
File "D:\Documents\Ripik\truck_detect\sort\sort.py", line 54, in iou_batch
xx1 = np.maximum(bb_test[..., 0], bb_gt[..., 0])
ValueError: operands could not be broadcast together with shapes (0,) (1,3)

[ZXStudio]

I have encountered similar problems before. Thank you for your answer at @sauravns. In the iou_batch() function of sort.py, we should first check if the length of bb_test is 0, and then set the detection to an empty numpy array.

def iou_batch(bb_test, bb_gt): if len(bb_test) == 0: bb_test = np.empty((0, 5))