However, we observe that the speed and accuracy of YOLOs are negatively affected by the NMS. Recently, end-to-end Transformer-based detectors (DETRs) have provided an alternative to eliminating NMS. Nevertheless, the high computational cost limits their practicality and hinders them from fully exploiting the advantage of excluding NMS. In this paper, we propose the Real-Time DEtection TRansformer (RT-DETR), the first real-time end-to-end object detector to our best knowledge that addresses the above dilemma. We build RT-DETR in two steps, drawing on the advanced DETR: first we focus on maintaining accuracy while improving speed, followed by maintaining speed while improving accuracy.…
Open-weight model · Object detection
dfine-medium-coco
by University of Science and Technology of China ustc-community/dfine-medium-coco
The D-FINE model was proposed in D-FINE: Redefine Regression Task in DETRs as Fine-grained Distribution Refinement by Yansong Peng, Hebei Li, Peixi Wu, Yueyi Zhang, Xiaoyan Sun, Feng Wu This model was contributed by VladOS95-cyber with the help of @qubvel-hf…
Parameters20M
Context—
Weights78.7 MB
Licenseapache-2.0
AccessOpen weights
Monthly Downloads6.9k
Runs On
What it takes to serve dfine-medium-coco (20M parameters): the memory its weights need at each precision, and the cheapest way to rent enough data-center GPUs to hold them.
| Precision | Weights | Memory needed | Cheapest setup | Per hour | Also fits |
|---|---|---|---|---|---|
| 16-bit | 0.0 GB | 0.0 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
| 8-bit | 0.0 GB | 0.0 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
| 4-bit | 0.0 GB | 0.0 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
Memory is the weights at that precision plus 20% for the runtime and a short context; a long context needs more. Prices are the lowest on-demand hourly rates in the SAVRN Index, read Sep 18, 2026.
Model Card
By University of Science and Technology of China, published under apache-2.0, revision 4ab9f5e46643.
The D-FINE model was proposed in D-FINE: Redefine Regression Task in DETRs as Fine-grained Distribution Refinement by Yansong Peng, Hebei Li, Peixi Wu, Yueyi Zhang, Xiaoyan Sun, Feng Wu This model was contributed by VladOS95-cyber with the help of @qubvel-hf This is the HF transformers implementation for D-FINE coco -> model trained on COCO obj365 -> model trained on Object365 obj2coco -> model trained on Object365 and then finetuned on COCO D-FINE, a powerful real-time object detector that achieves outstanding localization precision by redefining the bounding box regression task in DETR models. D-FINE comprises two key components: Fine-grained Distribution Refinement (FDR) and Global…
Read University of Science and Technology of China's full model card
D-FINE
Overview
The D-FINE model was proposed in D-FINE: Redefine Regression Task in DETRs as Fine-grained Distribution Refinement by Yansong Peng, Hebei Li, Peixi Wu, Yueyi Zhang, Xiaoyan Sun, Feng Wu
This model was contributed by VladOS95-cyber with the help of @qubvel-hf
This is the HF transformers implementation for D-FINE
_coco -> model trained on COCO
_obj365 -> model trained on Object365
_obj2coco -> model trained on Object365 and then finetuned on COCO
Performance
D-FINE, a powerful real-time object detector that achieves outstanding localization precision by redefining the bounding box regression task in DETR models. D-FINE comprises two key components: Fine-grained Distribution Refinement (FDR) and Global Optimal Localization Self-Distillation (GO-LSD).
How to use
import torch
import requests
from PIL import Image
from transformers import DFineForObjectDetection, AutoImageProcessor
url = 'http://images.cocodataset.org/val2017/000000039769.jpg'
image = Image.open(requests.get(url, stream=True).raw)
image_processor = AutoImageProcessor.from_pretrained("ustc-community/dfine-medium-coco")
model = DFineForObjectDetection.from_pretrained("ustc-community/dfine-medium-coco")
inputs = image_processor(images=image, return_tensors="pt")
with torch.no_grad():
outputs = model(**inputs)
results = image_processor.post_process_object_detection(outputs, target_sizes=torch.tensor([image.size[::-1]]), threshold=0.3)
for result in results:
for score, label_id, box in zip(result["scores"], result["labels"], result["boxes"]):
score, label = score.item(), label_id.item()
box = [round(i, 2) for i in box.tolist()]
print(f"{model.config.id2label[label]}: {score:.2f} {box}")
Training
D-FINE is trained on COCO (Lin et al. [2014]) train2017 and validated on COCO val2017 dataset. We report the standard AP metrics (averaged over uniformly sampled IoU thresholds ranging from 0.50 − 0.95 with a step size of 0.05), and APval5000 commonly used in real scenarios.
Applications
D-FINE is ideal for real-time object detection in diverse applications such as autonomous driving, surveillance systems, robotics, and retail analytics. Its enhanced flexibility and deployment-friendly design make it suitable for both edge devices and large-scale systems + ensures high accuracy and speed in dynamic, real-world environments.
Configuration
- Architecture
- DFineForObjectDetection
- Stored precision
- float32
- Model type
- d_fine
Identity and Version
- Repository
- ustc-community/dfine-medium-coco
- Publisher
- University of Science and Technology of China
- Task
- Object detection
- Modality
- Image
- Library
- transformers
- Parameters
- 20M parameters
- Languages
- en
- Revision
- 4ab9f5e466432f4fcf9a9a023da6aa5ecc9c9829
- First published
- 2025-02-22
- Last updated
- 2025-05-05
Files and Weights
5 files, 78.7 MB in total. The weights are 1 file totalling 78.7 MB in safetensors.
Weights1 file · 78.7 MB
Configuration2 files · 7.4 KB
Documentation1 file · 2.8 KB
Repository1 file · 1.5 KB
Every file
| File | Type | Size | SHA-256 |
|---|---|---|---|
| model.safetensors | Weights | 78.7 MB | 5a53154fb87a |
| config.json | Configuration | 7.0 KB | — |
| preprocessor_config.json | Configuration | 470 B | — |
| README.md | Documentation | 2.8 KB | — |
| .gitattributes | Repository | 1.5 KB | — |
License and Download
- License
- apache-2.0
- Access
- Open weights, no gate
- Download size
- 78.7 MB
Download from University of Science and Technology of China
Released by University of Science and Technology of China through its official repository on Hugging Face. Read the license.
Built From
- Described by arXiv:2410.13842
- Trained on (disclosed) coco
Memory Requirements
| Precision | Weights in memory |
|---|---|
| As published | 78.7 MB |
| 16-bit | 0.0 GB |
| 8-bit | 0.0 GB |
| 4-bit | 0.0 GB |
Weights only, from the published parameter count; the key-value cache and runtime add to this.
Questions About dfine-medium-coco
How much GPU memory does dfine-medium-coco need?
About 0 GB at 16-bit and 0 GB at 4-bit: the weights (20M parameters) plus a working margin. A long context needs more.
What is the cheapest GPU to run dfine-medium-coco on?
At 16-bit, 1x MI300X from $1.85 an hour; at 4-bit, 1x MI300X from $1.85 an hour, at the lowest on-demand prices the SAVRN Index lists.
Can I use dfine-medium-coco commercially?
Yes. dfine-medium-coco is released under Apache License 2.0. The Apache License 2.0 is a permissive open-source license. It permits commercial use, modification and redistribution. It requires keeping the license and copyright notices and any NOTICE file, stating significant changes, and it includes an express patent grant from contributors.
Similar Models
However, we observe that the speed and accuracy of YOLOs are negatively affected by the NMS. Recently, end-to-end Transformer-based detectors (DETRs) have provided an alternative to eliminating NMS. Nevertheless, the high computational cost limits their practicality and hinders them from fully exploiting the advantage of excluding NMS. In this paper, we propose the Real-Time DEtection TRansformer (RT-DETR), the first real-time end-to-end object detector to our best knowledge that addresses the above dilemma. We build RT-DETR in two steps, drawing on the advanced DETR: first we focus on maintaining accuracy while improving speed, followed by maintaining speed while improving accuracy.…
The RT-DETRv2 model was proposed in RT-DETRv2: Improved Baseline with Bag-of-Freebies for Real-Time Detection Transformer by Wenyu Lv, Yian Zhao, Qinyao Chang, Kui Huang, Guanzhong Wang, Yi Liu. RT-DETRv2 refines RT-DETR by introducing selective multi-scale feature extraction, a discrete sampling operator for broader deployment compatibility, and improved training strategies like dynamic data augmentation and scale-adaptive hyperparameters. These changes enhance flexibility and practicality while maintaining real-time performance. This model was contributed by @jadechoghari with the help of @cyrilvallez and @qubvel-hf This is RT-DETRv2 consistently outperforms its predecessor across all…
LW-DETR, a Light-Weight DEtection TRansformer model, is designed to be a real-time object detection alternative that outperforms conventional convolutional (YOLO-style) and earlier transformer-based (DETR) methods in terms of speed and accuracy trade-off. It was introduced in the paper LW-DETR: A Transformer Replacement to YOLO for Real-Time Detection by Chen et al. and first released in this repository. Disclaimer: This model was originally contributed by stevenbucaille in transformers. LW-DETR is an end-to-end object detection model that uses a Vision Transformer (ViT) backbone as its encoder, a simple convolutional projector, and a shallow DETR decoder. The core philosophy is to leverage…
Table Transformer (DETR) model trained on PubTables1M. It was introduced in the paper PubTables-1M: Towards Comprehensive Table Extraction From Unstructured Documents by Smock et al. and first released in this repository. Disclaimer: The team releasing Table Transformer did not write a model card for this model so this model card has been written by the Hugging Face team. The Table Transformer is equivalent to DETR, a Transformer-based object detection model. Note that the authors decided to use the "normalize before" setting of DETR, which means that layernorm is applied before self- and cross-attention. You can use the raw model for detecting tables in documents. See the documentation for…
Table Transformer (DETR) model trained on PubTables1M. It was introduced in the paper PubTables-1M: Towards Comprehensive Table Extraction From Unstructured Documents by Smock et al. and first released in this repository. Disclaimer: The team releasing Table Transformer did not write a model card for this model so this model card has been written by the Hugging Face team. The Table Transformer is equivalent to DETR, a Transformer-based object detection model. Note that the authors decided to use the "normalize before" setting of DETR, which means that layernorm is applied before self- and cross-attention. You can use the raw model for detecting the structure (like rows, columns) in tables.…