Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

FSAF on mmdet v2.0 #2520

Closed
wants to merge 12 commits into from
Closed

FSAF on mmdet v2.0 #2520

Changes from 1 commit
Commits
Show all changes
12 commits
Select commit Hold shift + click to select a range
6f174a6
starting to merge fsaf
Johnson-Wang Dec 27, 2019
f1ba30c
changed configs
Johnson-Wang Apr 22, 2020
32d49a1
changed bbox_coder
Johnson-Wang Apr 22, 2020
2290881
get_targets fun in AnchorHead class is modified to enable additional …
Johnson-Wang Apr 23, 2020
544c169
effective area assigner and tblr bbox_coder are implemented
Johnson-Wang Apr 23, 2020
827f164
added fsaf_head and config
Johnson-Wang Apr 23, 2020
c8e6244
added more fsaf configs
Johnson-Wang Apr 23, 2020
16e355c
fixed the nan bbox_loss by clamping bbox_pred with min=1e-4
Johnson-Wang Apr 23, 2020
fece78d
fixed the nan bbox_loss by clamping bbox_pred with min=1e-4
Johnson-Wang Apr 23, 2020
591e473
Merge remote-tracking branch 'remotes/origin/v2.0' into v2.0
Johnson-Wang Apr 23, 2020
4238ca2
changed config
Johnson-Wang Apr 23, 2020
d127d08
formatter
Johnson-Wang Apr 23, 2020
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
Prev Previous commit
Next Next commit
fixed the nan bbox_loss by clamping bbox_pred with min=1e-4
  • Loading branch information
@Johnson-Wang
Johnson-Wang committed Apr 23, 2020
commit fece78dd2cd4e8eb0d99a9b2714c00a12f1959b5
4 changes: 2 additions & 2 deletions mmdet/models/anchor_heads/fsaf_head.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -182,8 +182,8 @@ def loss(self,
all_anchor_list = images_to_levels(concat_anchor_list,
num_level_anchors)
for i in range(len(bbox_preds)):
bbox_preds[i].clamp_(
min=1e-4) # avoid 0 area of the predicted bbox
bbox_preds[i] = bbox_preds[i].clamp(min=1e-4)
# avoid 0 area of the predicted bbox
losses_cls, losses_bbox = multi_apply(
self.loss_single,
cls_scores,
Expand Down