rcd

DETR (7h3cyb3rm0nk/rcd) fine-tuned on RDD2022 for 4 road-damage classes.

Labels

id class
0 longitudinal_crack
1 transverse_crack
2 alligator_crack
3 pothole

Training

  • dataset: dronefreak/RDD2022 (5000 train / 800 val images)
  • epochs: 20, batch size: 4, lr: 5e-05, seed: 42
  • input size: shortest edge 640 / longest edge 1024
  • precision: bf16

Validation (subset of 800 images)

metric value
mAP@0.5 0.032
mAP@[.5:.95] 0.009

Uploaded as version rdd2022-detr-20ep-20261022-043741 on 2026-10-22 04:37 UTC.

Downloads last month
101
Safetensors
Model size
41.6M params
Tensor type
F32
ยท
Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ Ask for provider support

Model tree for 7h3cyb3rm0nk/rcd

Unable to build the model tree, the base model loops to the model itself. Learn more.

Dataset used to train 7h3cyb3rm0nk/rcd