THU-MIG Vision Models

Compare all 3 THU-MIG vision models we track, all of them open-weight. Try 1 of them on your own images, free in the Roboflow Playground.

3 models · 3 open-weight · 1 free to try · Updated Mar 2025

About THU-MIG Vision Models

Every THU-MIG vision model in our catalog is open-weight. All 3 publish downloadable weights, so you can self-host under their own licenses instead of paying per request. The open-weight side runs on AGPL 3.0 terms. The newest additions are YOLOE (Mar 2025) and YOLOv12 (Feb 2025).

Between them, the THU-MIG models we list cover 5 distinct vision tasks. The widest coverage is object detection (3 models), instance segmentation (2), and classification (1). For object detection, start with YOLOE, YOLOv12, or YOLOv10.

On the open-weight tier, YOLOv12 is the largest at 2.6M-59.1M parameters and YOLOv10 the smallest at 2.3M-29.5M, small enough to run on a single GPU or on-device. All 3 ship under custom or copyleft terms, so read the license before you deploy commercially.

With every model here open-weight, the real trade is size against generality: the smaller checkpoints fine-tune and deploy cheaply on your own hardware, while the larger ones cover more ground out of the box. 1 of the 3 THU-MIG vision models run live in the Roboflow Playground, so you can run the same image through several of them and compare the answers before committing to one.

Which THU-MIG Model Should You Use?

What each of the 3 THU-MIG vision models in our catalog is built for, and how you run it.

YOLOE: Best for: Instance Segmentation and Open Vocabulary Object Detection. Open weights: AGPL 3.0 license, 10M-50M parameters. Self-host it or run it here. Small enough for on-device and edge deployment.
YOLOv12: Best for: Instance Segmentation and Object Detection. Open weights: AGPL 3.0 license, 2.6M-59.1M parameters. Self-host it or run it here. Small enough for on-device and edge deployment.
YOLOv10: Best for: Object Detection. Open weights: AGPL 3.0 license, 2.3M-29.5M parameters. Self-host it or run it here. Small enough for on-device and edge deployment. Runnable in the Playground.

Open-Source THU-MIG Models

3 models with downloadable weights you can self-host under their licenses (AGPL 3.0). 1 run live in the Playground through hosted APIs, so self-hosting is optional.

YOLOE

YOLOE (YOLO with Everything) is an open-vocabulary object detection and segmentation model developed by THU-MIG at Tsinghua University, released in March 2025 under the AGPL-3.0 license. It extends the YOLO architecture to support open-vocabulary detection through text and visual prompts, enabling the model to detect arbitrary object categories beyond a fixed training set without retraining. The design integrates prompt encoding directly into the YOLO framework while preserving real-time inference speed.YOLOE is evaluated on COCO and LVIS benchmarks and supports both closed-set and open-vocabulary detection modes. It is built on the Ultralytics codebase and maintains compatibility with standard YOLO training and export workflows. YOLOE is suited for applications requiring flexible, prompt-driven object detection where the target object vocabulary may change at inference time.

YOLOv12

YOLOv12 is an attention-centric real-time object detection model developed by researchers at Tsinghua University, with the arXiv paper published in February 2025 under the AGPL-3.0 license. It introduces an Area Attention module that partitions feature maps into regions and applies self-attention within each region, reducing the quadratic complexity of full self-attention while capturing long-range dependencies. It also incorporates R-ELAN for improved feature aggregation and scaled residual connections for training stability.YOLOv12-L achieves 54.0% AP on COCO, while the YOLOv12-N variant achieves 40.5% mAP at 1.62ms latency on an NVIDIA T4 GPU. The model is built on the Ultralytics codebase, supporting detection, segmentation, and other standard YOLO tasks at competitive real-time speeds.

YOLOv10

YOLOv10 is a real-time end-to-end object detection model developed by THU-MIG at Tsinghua University, released in May 2024 under the AGPL-3.0 license. It introduces consistent dual assignments during training — using both one-to-many and one-to-one label assignment strategies — to eliminate the need for non-maximum suppression at inference time while maintaining competitive accuracy. This end-to-end design reduces inference latency compared to NMS-dependent detectors at similar accuracy levels.YOLOv10-B achieves 52.7% AP on COCO with 46% lower latency than YOLOv9-C at comparable performance. The model is available in six sizes from Nano to Extra Large, built on the Ultralytics framework, and exportable to ONNX, TensorRT, and CoreML. YOLOv10 is suited for latency-sensitive deployment scenarios where post-processing overhead is a constraint.

Frequently Asked Questions About THU-MIG Vision Models

Are THU-MIG vision models open source?

Yes. All 3 THU-MIG vision models we track publish downloadable weights you can self-host under their licenses (AGPL 3.0).

How many THU-MIG vision models are on Roboflow Playground?

We track 3 live THU-MIG vision models. The most recent addition is YOLOE, released Mar 2025.

Can I try THU-MIG vision models for free?

Yes. 1 of the 3 THU-MIG models run live in the Roboflow Playground. Upload your own image, run several models on it at once, and compare the outputs side by side. You can try it instantly, and a free account unlocks unlimited runs.

This page lists all 3 THU-MIG vision models in the Roboflow Playground catalog, all of them open-weight and free to self-host. They cover object detection, instance segmentation, and classification, among other tasks. 1 of them run live in the Roboflow Playground on your own images. Compare licenses, parameters, prices, and release dates side by side, or open any model page for full details.