---
license: apache-2.0
pipeline_tag: object-detection
library_name: rfdetr
datasets:
- dronefreak/UAVDT
tags:
- object-detection
- detectionbench
- rfdetr
- pytorch
- computer-vision
- aerial-imagery
- drone
- uav
- vehicle-detection
- traffic-surveillance
- small-object-detection
metrics:
- map50
- map50-95
- precision
- recall
- f1
base_model: "Roboflow/rf-detr-medium"
model-index:
- name: RF-DETR Medium Finetuned on UAVDT
results:
- task:
type: object-detection
name: Object Detection
dataset:
name: UAVDT
type: uavdt
metrics:
- type: mAP50
value: 33.28
name: mAP@50 (test split)
- type: mAP50-95
value: 20.54
name: mAP@50-95 (test split)
- type: precision
value: 73.03
name: Precision (test split)
- type: recall
value: 70.03
name: Recall (test split)
source:
url: https://github.com/dronefreak/DetectionBench
name: DetectionBench
---
# RF-DETR Medium Finetuned on UAVDT
Fine-tuned RF-DETR Medium object detector on the **UAVDT** benchmark dataset, trained and evaluated as part of [DetectionBench](https://github.com/dronefreak/DetectionBench) -- a framework for reproducibly benchmarking modern object detectors with identical training recipes and evaluation metrics across multiple real-world datasets.
---
## Usage
### Install Dependencies
```bash
pip install rfdetr huggingface_hub
```
### Load Model from Hugging Face
```python
from huggingface_hub import hf_hub_download
import rfdetr
weights = hf_hub_download(
repo_id="dronefreak/uavdt-rfdetr-medium",
filename="checkpoint_best_total.pth"
)
model = rfdetr.RFDETRMedium(pretrain_weights=weights)
```
### Run Inference
```python
detections = model.predict("image.jpg", threshold=0.25)
```
---
## Performance
Evaluated on the UAVDT **test** split, using DetectionBench's standard evaluation pipeline (`detectionbench-evaluate`).
| Metric | Score (%) |
| ---------- | --------------- |
| mAP@50 | 33.28 |
| mAP@50-95 | 20.54 |
| Precision | 73.03 |
| Recall | 70.03 |
| F1 Score | 71.5 |
| Parameters | 33.7M |
| FLOPs | N/A (not published upstream) |
---
## UAVDT Model Zoo
Every model DetectionBench has trained and evaluated on UAVDT so far, for full transparency -- see [DetectionBench](https://github.com/dronefreak/DetectionBench) for the smaller, curated comparison set used on the project README.
| Model | mAP@50 | mAP@50-95 | Precision | Recall |
| --------------------- | ------------- | --------------- | ----------------- | -------------- |
| YOLO26m | 33.43 | 19.56 | 38.14 | 39.84 |
| **RF-DETR Medium** | **33.28** | **20.54** | **73.03** | **70.03** |
| YOLO26s | 32.98 | 19.61 | 43.86 | 40.38 |
| RF-DETR Nano | 32.78 | 20.31 | 73.6 | 66.98 |
| YOLO26x | 32.65 | 19.22 | 41.85 | 38.19 |
| YOLO26l | 32.64 | 18.75 | 40.17 | 36.45 |
| RF-DETR Small | 32.62 | 20.21 | 73.83 | 71.63 |
| YOLOv9s | 31.82 | 18.71 | 39.83 | 38.12 |
| YOLOv8m | 31.42 | 18.8 | 40.27 | 37.79 |
| YOLO11x | 31.05 | 18.31 | 37.4 | 36.38 |
| YOLO11m | 30.47 | 17.71 | 37.7 | 37.01 |
| YOLOv8x | 30.47 | 17.66 | 39.61 | 36.16 |
| YOLOv10m | 30.12 | 17.33 | 40.13 | 35.68 |
| YOLOv9m | 29.43 | 16.97 | 35.92 | 35.7 |
| YOLOv9t | 29.42 | 17.03 | 35.75 | 36.47 |
| YOLOv10x | 29.38 | 17.15 | 37.29 | 35.15 |
| YOLOv10l | 29.16 | 16.54 | 36.9 | 35.6 |
| YOLOv9c | 29.16 | 16.46 | 35.38 | 34.35 |
| YOLO11s | 29.1 | 17.16 | 34.32 | 37.31 |
| YOLO26n | 28.88 | 16.79 | 33.14 | 35.66 |
| YOLOv8l | 28.86 | 17.27 | 38.33 | 32.86 |
| YOLOv10s | 28.85 | 16.48 | 36.53 | 33.16 |
| YOLO11l | 28.64 | 17.16 | 34.75 | 34.02 |
| YOLO11n | 28.56 | 16.3 | 38.04 | 32.26 |
| YOLOv8n | 27.8 | 15.34 | 35.42 | 33.61 |
| YOLOv10n | 27.17 | 15.16 | 33.3 | 31.21 |
| YOLOv8s | 27.12 | 15.33 | 34.65 | 31.87 |
---
## Per-Class Performance
| Class | mAP@50 | mAP@50-95 |
| -------------------------- | --------------- | ----------------- |
| car | 74.75 | 43.64 |
| truck | 6.73 | 4.51 |
| bus | 18.37 | 13.46 |
This model was evaluated with [Supervision](https://github.com/roboflow/supervision)'s detection metrics, which report mAP/Precision/Recall directly but don't produce a confusion-matrix plot the way Ultralytics' validator does.
---
## Dataset
This model was trained on **UAVDT**. For the full dataset description, provenance, license, and citation, see the dataset card:
https://huggingface.co/datasets/dronefreak/UAVDT
### Classes
* car
* truck
* bus
---
## Training Configuration
| Setting | Value |
| ---------------- | -------------------------------- |
| Dataset | UAVDT |
| Framework | RF-DETR |
| Training Toolkit | DetectionBench |
| Epochs (configured max) | 20 |
| Epochs (actually trained) | 0 |
| Early Stopping Patience | 5 |
| Batch Size | 12 |
| Resolution | 640 |
| Optimizer | adamw |
| Learning Rate | 5e-05 |
| Seed | 42 |
---
## Repository Contents
```text
checkpoint_best_total.pth
metrics.csv
config.json
uavdt_rfdetr-medium_showcase.jpg
assets/demo_banner.mp4
assets/demo_banner_poster.jpg
README.md
```
---
## Related Resources
* [UAVDT dataset card](https://huggingface.co/datasets/dronefreak/UAVDT) on Hugging Face
* [DetectionBench](https://github.com/dronefreak/DetectionBench) -- reproducible benchmarks for modern object detectors on real-world datasets
* [UAVDT paper preprint (arXiv:1804.00518)](https://arxiv.org/abs/1804.00518)
* [UAVDT project website (official data source)](https://sites.google.com/view/grli-uavdt)
---
## Training Framework
This model was trained using [DetectionBench](https://github.com/dronefreak/DetectionBench), an open-source framework for benchmarking object detectors across multiple real-world datasets with a common pipeline.
Features include:
* A dataset-adapter registry for converting real-world datasets into a canonical format
* Identical training/evaluation recipes across model families (Ultralytics YOLO/RT-DETR, RF-DETR)
* Hardware profiling (latency, FPS, VRAM, parameters, FLOPs)
* One-command reproducibility via versioned Hydra configs
If you find this model useful, please consider starring the repository.
---
## Known Limitations
* Severe class imbalance: `car` (94.6%) dominates the annotated boxes, while `truck` (3.1%) and `bus` (2.3%) are rare -- per-class accuracy on the minority classes is measured on comparatively few examples, and every model here scores far lower on them than on `car`.
* Very small objects: the median box covers only 0.14% of the image area (mean 0.26%), so this is a hard small-object regime and absolute mAP values are low for every architecture; the numbers are best read as a relative comparison between models, not as a production-quality detector.
* Video-derived, highly correlated frames: the ~40.7k labelled images come from 50 video sequences, so consecutive frames are near-duplicates. UAVDT's 50 tracking-only sequences have no detection labels and are excluded. The validation split is carved out of the training sequences by sequence (not by frame) to avoid leakage, but effective diversity is far lower than the image count suggests.
* Different density per split: instances per image are 15.7 (train), 28.0 (valid) and 22.7 (test), because the splits contain different sequences -- validation metrics are not directly predictive of test metrics.
* Research-use-only data: UAVDT is distributed "for research purpose only" with no redistribution grant, so the dataset is not mirrored here -- obtain it from the official source (see the Dataset section above) and check its terms before any use beyond research.
---
## Citation
If you use this model in your research, please consider citing the dataset and the model architecture:
```
@InProceedings{du2018unmanned,
title={The Unmanned Aerial Vehicle Benchmark: Object Detection and Tracking},
author={Du, Dawei and Qi, Yuankai and Yu, Hongyang and Yang, Yifan and Duan, Kaiwen and Li, Guorong and Zhang, Weigang and Huang, Qingming and Tian, Qi},
booktitle={Proceedings of the European Conference on Computer Vision (ECCV)},
year={2018}
}
```
```bibtex
@inproceedings{robinson2026rfdetr,
title = {RF-DETR: Real-Time Detection Transformer},
author = {Robinson, Isaac and Robicheaux, Peter and Popov, Matvei and Ramanan, Deva and Peri, Neehar},
booktitle = {International Conference on Learning Representations (ICLR)},
year = {2026},
url = {https://arxiv.org/abs/2511.09554}
}
@article{oquab2023dinov2,
title={DINOv2: Learning Robust Visual Features without Supervision},
author={Oquab, Maxime and Darcet, Timoth{\'e}e and Moutakanni, Theo and Vo, Huy and Szafraniec, Marc and Khalidov, Vasil and Fernandez, Pierre and Haziza, Daniel and Massa, Francisco and El-Nouby, Alaaeldin and others},
journal={arXiv preprint arXiv:2304.07193},
year={2023}
}
```
