README.md

---
license: apache-2.0
library_name: rfdetr.cpp
tags:
  - object-detection
  - rfdetr
  - gguf
  - ggml
  - cpp-inference
pipeline_tag: object-detection
base_model: roboflow/rfdetr
---

# RF-DETR Medium — GGUF for rfdetr.cpp

GGUF-format weights of [Roboflow RF-DETR Medium](https://github.com/roboflow/rf-detr) (detection variant) for use with [rfdetr.cpp](https://github.com/adithyab94/rf-detr.cpp), a C++/ggml implementation that matches the upstream PyTorch model on CPU.

This repo contains all four standard quantizations of this variant. **F16 is the recommended default** — same accuracy as F32, 1.85× smaller, and typically the fastest on modern CPUs thanks to ggml's F32×F16 matmul fast path.

## Available files

| File | Quant | Size (MB) | Recall @ IoU 0.5 | Recall @ IoU 0.95 | Mean \|Δscore\| | Latency (median ms, T=8) |
|---|---|---:|---:|---:|---:|---:|
| `rfdetr-medium-f32.gguf` | F32 | 125.0 | 0.9731 | 0.9621 | 0.0024 | 139.9 |
| `rfdetr-medium-f16.gguf` ← **recommended** | F16 | 67.2 | 0.9780 | 0.9670 | 0.0022 | 143.9 |
| `rfdetr-medium-q8_0.gguf` | Q8_0 | 40.2 | 0.9890 | 0.9691 | 0.0070 | 145.8 |
| `rfdetr-medium-q4_K.gguf` | Q4_K | 32.5 | 0.8828 | 0.7390 | 0.0199 | 155.8 |

All accuracy numbers above are computed against the upstream PyTorch reference (`rfdetr 1.9.0`) on 7 images (000000000139.jpg, 000000000632.jpg, 000000039769.jpg, 000000087038.jpg, 000000252219.jpg, 000000397133.jpg, bus.jpg) at threshold 0.5. Latency is measured separately with `rfdetr-cli bench` (8 iters + 3 warmup) at T=8 threads on a single Intel Core i7-12800HX, on `tests/fixtures/ci/test_image.jpg`.

## Architecture

- Backbone: DINOv2-small
- Input resolution: 576×576
- Patch size: 16
- Decoder layers: 4
- Object queries: 300
- Task: object detection (boxes only)

## Quantization notes

- **F32** — full-precision reference, ~120 MB. Bit-exact PyTorch parity.
- **F16** — matmul-multiplicand weights only; LayerNorms, conv kernels, embeddings, biases, and layer-scale gammas stay F32. Lossless on this model and consistently the fastest variant on CPU.
- **Q8_0** — best size/accuracy tradeoff under F16; ~3× smaller than F32 with effectively identical detections.
- **Q4_K** — smallest practical quant. Rows with `ne[0] % 256 != 0` (the decoder's 128-dim MLP halves, 60 tensors) silently fall back to Q8_0 per ggml's quantizer logic — net compression is still ~3.8× over F32. Use only when the size budget is tight; expect a measurable Recall@0.95 drop relative to F16/Q8_0 (see file table above).

## Compatibility

These GGUFs stamp `rfdetr.preprocess.resize_mode = "bilinear_no_antialias"`, matching RF-DETR 1.9's antialias-free float bilinear resize (`align_corners=false`, half-pixel coordinates, no intermediate uint8 rounding). rf-detr.cpp treats this key as **optional**: GGUFs that predate this metadata (no `resize_mode` key) keep using the legacy stb-based resize path, so older files continue to produce their original outputs unchanged. An unrecognized `resize_mode` value is rejected rather than guessed.

**Keypoint-preview inference is not supported.** rf-detr.cpp does not implement the keypoint output head; this repository only serves box detection outputs.

## Usage

```bash
# 1. Clone + build rfdetr.cpp
git clone https://github.com/adithyab94/rf-detr.cpp
cd rf-detr.cpp
cmake -B build -DRFDETR_BUILD_CLI=ON && cmake --build build -j

# 2. Download a quant (F16 recommended)
hf download adithya-balaji/rfdetr-cpp-medium rfdetr-medium-f16.gguf --local-dir models/

# 3. Run detection
build/bin/rfdetr-cli detect \
    --model models/rfdetr-medium-f16.gguf \
    --input my_image.jpg \
    --threshold 0.5 --threads 8 \
    --output detections.json
```

## Accuracy methodology

All accuracy metrics are computed against the upstream PyTorch reference (`rfdetr 1.9.0`) on 7 images (000000000139.jpg, 000000000632.jpg, 000000039769.jpg, 000000087038.jpg, 000000252219.jpg, 000000397133.jpg, bus.jpg) at threshold 0.5. Each detection match uses greedy Hungarian-style assignment by IoU (≥ 0.5 lenient, ≥ 0.95 strict) with class equality required.

See [BENCHMARK.md](https://github.com/adithyab94/rf-detr.cpp/blob/main/BENCHMARK.md) and [`benchmarks/results/accuracy_sweep.json`](https://github.com/adithyab94/rf-detr.cpp/blob/main/benchmarks/results/accuracy_sweep.json) for the full sweep across the (variant × quant) cells.

## Provenance

- Source project: [Roboflow RF-DETR](https://github.com/roboflow/rf-detr)
- Upstream package: `rfdetr==1.9.0`
- Converted with [rfdetr.cpp](https://github.com/adithyab94/rf-detr.cpp) at commit [`fbef9387bed3`](https://github.com/adithyab94/rf-detr.cpp/commit/fbef9387bed3f3c82bf3f4e041253ffe3081f275)
- Checkpoint: official pretrained `rfdetr-medium` weights (downloaded by the `rfdetr` package on first use)

### Checksums (SHA-256)

Also available as `SHA256SUMS` in this repo.

```
664ffe389100cbc942a5d99a2d5b48ef12fdf88f047471fd42d98c5c7d8d32fb  rfdetr-medium-f32.gguf
353bdccb3a6aa4865fe5c6b636d627c7d6f8c6f94701b9d074d5c2b757b4c792  rfdetr-medium-f16.gguf
b8f77932ceb8429856588aabfa415a5a752eba2d6c212b043faf473a8aab575f  rfdetr-medium-q8_0.gguf
7a6c0da4b1b1070986f30582eb8b1b7439b386b15a2171707346eb7c402cca60  rfdetr-medium-q4_K.gguf
```

## License

Apache-2.0 — matches the upstream [rfdetr](https://github.com/roboflow/rf-detr) license.