rf-detr · Issues· 82 open
Open on GitHubLocally synced open issues (discussions stay on GitHub)
- #1410
Investigation: faster training throughput without accuracy regression — starting with `torch.compile`
enhancementUpdated Sep 18, 2026 - #1484
Clarify and align container-level p semantics for nested Albumentations transforms
bugquestionUpdated Sep 18, 2026 - #1414
Issue with drawing overlapping bounding boxes during the prediction phase after training a multiclass detection using RF-DETR
Updated Sep 17, 2026 - #1441
Consider an opt-in ultrafast COCO metric backend: 2.09x faster with 37% less peak RSS
Updated Sep 8, 2026 - #1272
Query embedding extraction
enhancementUpdated Sep 3, 2026 - #1133
Proposal: Gather exceptions and name them
enhancementUpdated Aug 26, 2026 - #1347
Suggestion: merge dataset in `val` and `train` folder to train for final epochs
enhancementUpdated Aug 25, 2026 - #1058
TPU Training Support
enhancementUpdated Aug 23, 2026 - #1344
Suggestion: store decoded dataset images in RAM one time only instead of before every epoch
enhancementUpdated Aug 18, 2026 - #1153
The slow progress of RF-DETR training and the post-training validation process
bugUpdated Aug 17, 2026 - #1354
Support training on a selected subset of classes
questionUpdated Aug 17, 2026 - #1173
Image Classification model
questionUpdated Aug 14, 2026 - #1252
Kornia backend rejects 12 transform documented in `aug_configs`
bugUpdated Aug 11, 2026 - #1286
Weighted multi-source batch sampling: fixed per-source ratio in every training batch
enhancementUpdated Aug 5, 2026 - #780
RF-DETR Seg + SAHI: still missing cells in dense tiles, and inference is too slow (1900–2200px) — tuning & speedup advice?
enhancementUpdated Aug 3, 2026