mirrors
/
yolov5
mirror of https://github.com/ultralytics/yolov5.git
1
0
Fork 0
Code Issues Projects Releases Wiki Activity

Increase `ar_thr` from 20 to 100 for better detection on slender (high aspect ratio) objects (#5556) 

* Making `ar_thr` available as a hyperparameter

* Disabling ar_thr as hyperparameter and computing from the dataset instead

* Fixing bug in ar_thr computation

* Fix `ar_thr` to 100
pull/5993/head
Mrinal Jain 2021-12-15 19:12:23 +05:30 committed by GitHub
parent c9a46a60e0
commit b7d18f3636
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
1 changed files with 1 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
utils/augmentations.py
View File

@ -269,7 +269,7 @@ def mixup(im, labels, im2, labels2):
return im, labels
def box_candidates(box1, box2, wh_thr=2, ar_thr=20, area_thr=0.1, eps=1e-16): # box1(4,n), box2(4,n)
def box_candidates(box1, box2, wh_thr=2, ar_thr=100, area_thr=0.1, eps=1e-16): # box1(4,n), box2(4,n)
# Compute candidate boxes: box1 before augment, box2 after augment, wh_thr (pixels), aspect_ratio_thr, area_ratio
w1, h1 = box1[2] - box1[0], box1[3] - box1[1]
w2, h2 = box2[2] - box2[0], box2[3] - box2[1]