AnnotateIt/rtdetr-v2-r18vd-coco-onnx

RT-DETRv2 r18vd COCO — verified ONNX mirror for AnnotateIt

Byte-identical mirror of the FP32 ONNX graph from onnx-community/rtdetr_v2_r18vd-ONNX (revision 936f90b6a476c6da4dfe053fc521af55285976ba), based on the official Apache-2.0 PekingU/rtdetr_v2_r18vd checkpoint. AnnotateIt did not alter or quantize the graph; this mirror adds an explicit license, immutable provenance, checksums and independent validation. Not an official PekingU / RT-DETR release.

Artifact

File	Value
ONNX graph	model.onnx
Precision	FP32
Opset	16
Declared input	pixel_values: float32 [batch, 3, 640, 640]
Outputs	logits: float32 [batch, 300, 80]; pred_boxes: float32 [batch, 300, 4]
Size	81,057,510 bytes
SHA-256	583a236ac21c95a7fd94f284fc21485e42355bfef82c27011ba78fbc09ee87e2

Preprocessing and outputs

1. Convert to RGB.
2. Resize to 640×640, bilinear, stretch (no letterbox).
3. Rescale to [0, 1] (÷255); do not apply mean/std normalization.
4. HWC→NCHW, add batch dim.
5. sigmoid(logits), best COCO class per query; pred_boxes = normalized cxcywh → map to original size.

80 COCO classes, end-to-end set prediction — no NMS.

License

Apache-2.0 (see LICENSE); retain attribution when redistributing. Official project: lyuwenyu/RT-DETR.