Method Detail: DeepLabV3Plus-R34

Back to Leaderboard
Benchmark: UAV-Crack
Short name: DeepLabV3Plus-R34
Long name: DeepLabV3+ with ResNet34 for UAV Crack Segmentation
Description: DeepLabV3+ semantic segmentation model with a ResNet32 encoder pretrained on the ImageNet dataset. For binary UAV pavement crack segmentation, the model was trained with BSEWithLogits and Dice loss.Predictions were made using a threshold of 0.50
Reference: N/A
Last submitted: September 19, 2026
Published: September 19, 2026 at 19:08:01
Submissions: 1
Project page / code: N/A
Open source: No

Benchmark performance

Submission Date mIoU (↑) Crack F1 (↑) Crack IoU (↑) Crack Precision (↑) Crack Recall (↑) aAcc (↑)
2026-09-19 19:14 78.9400 75.7100 60.9200 74.5700 76.8900 97.1000