Bienqq
/
car-detection-bayes
2
1
Fork 0
Code Issues Pull Requests Releases Wiki Activity

updates

This commit is contained in:
Glenn Jocher 2019-02-18 19:58:01 +01:00
parent d81838e286
commit ce4ee36ca0
1 changed files with 3 additions and 8 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

11
utils/utils.py
View File

@ -345,7 +345,7 @@ def non_max_suppression(prediction, conf_thres=0.5, nms_thres=0.4):
class_prob, class_pred = torch.max(F.softmax(pred[:, 5:], 1), 1)
v = ((pred[:, 4] > conf_thres) & (class_prob > .4)) # TODO examine arbitrary 0.3 thres here
v = ((pred[:, 4] > conf_thres) & (class_prob > .4)) # TODO examine arbitrary 0.4 thres here
v = v.nonzero().squeeze()
if len(v.shape) == 0:
v = v.unsqueeze(0)
@ -388,9 +388,7 @@ def non_max_suppression(prediction, conf_thres=0.5, nms_thres=0.4):
dc = dc[1:][iou < nms_thres] # remove ious > threshold
# Image Total P R mAP
# 5000 5000 0.627 0.593 0.584
# 4964 5000 0.629 0.594 0.586 # complete probability sort
# 4964 5000 0.629 0.594 0.586
elif nms_style == 'AND': # requires overlap, single boxes erased
while len(dc) > 1:
@ -410,10 +408,7 @@ def non_max_suppression(prediction, conf_thres=0.5, nms_thres=0.4):
dc = dc[iou < nms_thres]
# Image Total P R mAP
# 4964 5000 0.632 0.597 0.588 # normal
# 4964 5000 0.632 0.597 0.588 # squared
# 4964 5000 0.631 0.597 0.588 # sqrt
# normal best_v1_0.pt
# 4964 5000 0.633 0.598 0.589 # normal
if len(det_max) > 0:
det_max = torch.cat(det_max)