Changelog

Since 1.4.0

Summarized from CHANGELOG.md on the dev branch, which carries the full entries.

New model families

Dome-DETR (domedetr) is the trainable addition: a tiny-object detector for aerial, drone and remote-sensing imagery, ported from the Apache-2.0 Dome-DETR release. It is D-FINE plus a density head (DeFE), encoder attention restricted to occupied windows (MWAS), and a query count set by local density rather than a fixed 300 (PAQI). Sizes s, m and l at 800x800, with a maximum absolute difference of 0.0 against upstream on all six published checkpoints. Its advantage narrows as objects grow, so it sits beside D-FINE rather than replacing it. There is no COCO checkpoint upstream, only AI-TOD-V2 with 9 classes and VisDrone with 12, so every canonical filename carries a dataset suffix. Export raises instead of emitting a graph, because PAQI's per-image query count makes a traced graph valid only for the image it was traced on. The code is Apache-2.0, but the weights are not rehosted: the upstream model card states no license in its metadata while its prose claims Apache-2.0 and restricts use to academic research.

The rest are inference-only ports of earlier published work, each checked against its pinned upstream source:

• Detection: DETR, Deformable DETR, DINO-DETR, LW-DETR, Faster R-CNN, Mask R-CNN, FCOS, RetinaNet, SSD300, CenterNet and EfficientDet.
• Classification: AlexNet, VGG, ViT, DeiT and Swin.
• Semantic segmentation: FCN and DeepLabv3.
• Depth: MiDaS. Pose: HRNet, top-down on COCO-17 person crops.

TEED and DexiNed arrived with the new edge task, as native MIT-licensed architectures with local checkpoint converters. Their upstream BIPED-trained checkpoints are not bundled, mirrored or auto-downloaded, because that dataset's terms are non-commercial.

LibreMODUS adds 14B-A7B analysis-only inference for depth, normals, edges, COCO detection and phrase grounding, with image-conditioned any2any() chaining. LibreFeyNobg adds a matte family at 1024 px, code and weights Apache-2.0.

New tasks and outputs

The edge and normal dense-prediction task contracts landed: original-canvas result payloads and visualization, dataset schemas, validators with edge ODS/OIS and normal angular metrics, and public API aliases.

RT-DETRv2 gained oriented-object detection inference for the official DOTA 1.0 checkpoints in sizes n, s, m, l and x, with aspect-preserving preprocessing, native Results.obb output, validation, and validated ONNX and TorchScript export.

Speed

CUDA graph capture of the training step went from 2 families to 24, across detect, classify, semantic, point and restore. Measured on an RTX 5070 Ti under AMP: 3.63x on FOMO, 2.74x on MobileNetV4, 1.99x on YOLO9-t, and 1.04x to 1.26x for the rest, with the win tracking how much of a step is network rather than loss. End to end on a 20-epoch YOLO9-t fine-tune of 406 images, dataloader and validation included, 428 s became 368 s with identical mAP50-95 and per-epoch losses.

Capture at prediction time went from 8 families to 39, spanning detect, segment, pose, point, classify, semantic, depth, restore, matte and OCR. Every enabled family is verified to replay bit-identically against two probe inputs. A family that cannot be captured whole is split at a verified seam, and the remainder runs eagerly with identical numbers.

pip install libreyolo[hub-kernels] opts every Deformable-DETR-lineage family into a compiled Apache-2.0 CUDA kernel for multi-scale deformable attention, pinned to an audited revision. It applies to eager CUDA fp32 only; exports keep the portable path.

Fused scaled dot product attention now runs across the transformer families using stock torch, with no optional dependency. Measured on an RTX 5070 Ti under fp16 autocast, roughly 2x on Swin window attention and 3.8x on OWLv2 vision attention. Families held to a byte-exact parity bar keep manual attention by default and opt in explicitly.

COCO metrics moved to faster-coco-eval by default: 15.6x faster overall and 56x on detection-dense datasets, decided on measured parity across all 100 RF100-VL test splits, where 1381 of 1400 metric values were bit-identical to pycocotools, the maximum deviation was 2.22e-16, and headline deltas were exactly 0. --no-faster-coco-eval opts out, and pycocotools stays the automatic fallback when the package is missing.

Training and validation

val_loss=True reached every trainable family, across detect, classify, semantic and restore rather than detection alone. It moved from YOLO9Config and RFDETRConfig to TrainConfig, so a family that has not implemented it now raises a clear error instead of ignoring the flag. Denoising terms are never included, because validation forwards without ground truth.

Comet, ClearML, Neptune and DVCLive joined the built-in training loggers, with the same canonical metrics and failure-isolation contract as the existing TensorBoard, MLflow and Weights & Biases integrations.

The kernel registry moved to libreyolo/kernels/, organized by purpose. LIBREYOLO_KERNELS replaces LIBREYOLO_QUANT_KERNELS, which is still honored, and libreyolo.quant.kernels remains a working alias.

Fixes

train(..., cuda_graph=True) had been a silent no-op for D-FINE, DEIM, DEIMv2, RT-DETRv4 and EC, whose trainers called the eager forward instead of the routed one, so capture never engaged. The end-to-end suite now asserts that capture actually engages per family.

YOLOX applies its BatchNorm eps=1e-3 and momentum=0.03 at construction instead of as a fixup afterwards, so the values survive the class-count rebuild that train() performs when a dataset's class count differs from the checkpoint. A fine-tune previously trained and reported in-training validation at torch's default eps=1e-5 but was reloaded for inference at 1e-3. On RF100-VL ball, the same nano checkpoint scores 0.566 mAP50-95 evaluated at its trained eps and 0.151 after a stock reload. Checkpoints trained before this fix carry the old semantics and need the override to report faithful numbers.

Non-strict checkpoint loads now warn with the counts and first names of missing and unexpected state-dict keys. Shape mismatches always raised, but name mismatches let a partially matching checkpoint load and then predict with freshly initialized tensors, leaving no trace.

CUDA graph capture no longer races with DataLoader pin-memory threads, which had killed runs with a capture error from the pin-memory thread. D-FINE training now applies upstream's per-size multi-scale recipe instead of a hardcoded base_size_repeat=3.

1.4.0, released 2026-07-24

The release notes summarize it as 15 new model families, 3 new tasks, a quantization stack, two new trackers, and a multi-GPU training correctness overhaul.

The new families were SegFormer, SwinIR, Real-ESRGAN, BiRefNet, ZipDepth, Depth Anything 3, PP-OCRv5, SigLIP 2, YOLOv1, SAM 3, EdgeTAM, PicoSAM3, OMDet-Turbo, OV-DEIM and SenseNova-Vision. The new tasks were panoptic, matte and ocr, each with its result type and validator, and EoMT gained instance and panoptic segmentation.

Quantization arrived as the libreyolo quantize command and model.quantize(), with fp16, bf16, fp8, int8, w4a16, w4a8, nvfp4 and mxfp4 recipes, and quantization-aware training through train() on a quantized checkpoint, for yolo9 and rfdetr. Tracking gained BoT-SORT and Deep OC-SORT with an OSNet-AIN re-identification embedder.

The multi-GPU work was about correctness. DDP now shards correctly for DEIM, D-FINE and YOLO-NAS-pose, where every rank had been training the full dataset at the full batch, and loss normalizers are globally all-reduced to match single-GPU gradients. SyncBatchNorm defaults on for the CNN detectors, and a non-divisible global batch or a custom loader that does not shard now raises at setup instead of running wrong.

Several training fixes changed results outright: RTMDet fine-tune collapse from missing head init took an nc=1 rebuild from 0.26 to 0.709 mAP50-95, YOLOv7 training color-space and fp16 overflow bugs took a fine-tune from 0.0 to 0.92 mAP50, and PicoDet's and DEIM's learning-rate defaults were lowered because the old ones destroyed pretrained weights. Segmentation training no longer exhausts RAM on COCO-scale datasets with multiple workers.

One compatibility note from that release: checkpoints written with the new task strings or with finalized quantization state cannot be loaded by 1.3.1.

Earlier releases

Releases before 1.4.0 are documented in the GitHub releases only, not in CHANGELOG.md. Their documentation is still online, listed on versions.