Supported Models

This page lists all models currently supported by trtutils.

Object Detection

YOLO Models

1. Ultralytics YOLO - YOLOv3, YOLOv5, YOLOv8, YOLOv11, and YOLOv26

   • YOLOv3: yolov3tu, yolov3u, yolov3sppu

   • YOLOv5: yolov5nu, yolov5su, yolov5mu, yolov5lu, yolov5xu (and the 6u variants)

   • YOLOv8: yolov8n, yolov8s, yolov8m, yolov8l, yolov8x

   • YOLOv11: yolov11n, yolov11s, yolov11m, yolov11l, yolov11x

   • YOLOv26: yolov26n, yolov26s, yolov26m, yolov26l, yolov26x

   • GitHub: ultralytics/ultralytics

2. YOLOv7

   • GitHub: WongKinYiu/yolov7

3. YOLOv9

   • GitHub: WongKinYiu/yolov9

4. YOLOv10

   • GitHub: THU-MIG/yolov10

5. YOLOv12

   • GitHub: sunsmarterjie/yolov12

6. YOLOv13

   • GitHub: iMoonLab/yolov13

7. YOLOX

   • GitHub: Megvii-BaseDetection/YOLOX

DETR Models

8. RT-DETRv1

   • GitHub: lyuwenyu/RT-DETR

9. RT-DETRv2

   • GitHub: lyuwenyu/RT-DETR

10. RT-DETRv3

    • GitHub: clxia12/RT-DETRv3

11. D-FINE

    • GitHub: Peterande/D-FINE

12. DEIM

    • GitHub: Intellindust-AI-Lab/DEIM

13. DEIMv2

    • GitHub: Intellindust-AI-Lab/DEIMv2

14. RF-DETR

    • GitHub: roboflow/rf-detr

Depth Estimation

1. Depth-Anything-V1

   • GitHub: LiheYoung/Depth-Anything

2. Depth-Anything-V2

   • GitHub: DepthAnything/Depth-Anything-V2

3. Depth-Anything-V3

   • GitHub: ByteDance-Seed/depth-anything-3

Classification

1. Torchvision Classifiers

   • See: PyTorch Vision Classification Models

Hand-Object Interaction

1. HOI-DETR

   • GitHub: AhmadDarKhalil/HOI-DETR

2. Hands23

   • GitHub: EvaCheng-cty/hands23_detector

Model Download Support

The following models can be automatically downloaded and converted to ONNX format using the download CLI command:

YOLO Models

• YOLOv3: yolov3tu, yolov3u, yolov3sppu (via Ultralytics)

• YOLOv5: yolov5nu, yolov5su, yolov5mu, yolov5lu, yolov5xu and the 6u variants (via Ultralytics)

• YOLOv7: All variants with pretrained weights

• YOLOv8: yolov8n, yolov8s, yolov8m, yolov8l, yolov8x (via Ultralytics)

• YOLOv9: All variants with pretrained weights

• YOLOv10: All variants with pretrained weights

• YOLOv11: yolov11n, yolov11s, yolov11m, yolov11l, yolov11x (via Ultralytics)

• YOLOv12: All variants with pretrained weights

• YOLOv13: yolov13n, yolov13s, yolov13l, yolov13x

• YOLOv26: yolov26n, yolov26s, yolov26m, yolov26l, yolov26x (via Ultralytics)

• YOLOX: yoloxn, yoloxt, yoloxs, yoloxm, yoloxl, yoloxx, yolox_darknet

DETR Models

• RT-DETRv1: Multiple configurations available

• RT-DETRv2: Multiple configurations available

• RT-DETRv3: Multiple configurations available

• D-FINE: Multiple configurations available

• DEIM: Multiple configurations available

• DEIMv2: deimv2_atto, deimv2_femto, deimv2_pico, deimv2_n, deimv2_s, deimv2_m, deimv2_l, deimv2_x

• RF-DETR: Multiple configurations available

Depth Estimation Models

• Depth-Anything-V1: depth_anything_v1_small, depth_anything_v1_base, depth_anything_v1_large

• Depth-Anything-V2: depth_anything_v2_small, depth_anything_v2_base, depth_anything_v2_large

• Depth-Anything-V3: depth_anything_v3_mono_large, depth_anything_v3_metric_large

Hand-Object Interaction Models

• HOI-DETR: hoi_detr_vitl

• Hands23: hands23_x101

Example Usage

Download models using the CLI:

$ python -m trtutils download --model yolov8n --output yolov8n.onnx
$ python -m trtutils download --model yolov11m --output yolov11m.onnx --imgsz 640 --opset 17
$ python -m trtutils download --model yoloxs --output yoloxs.onnx --imgsz 640 --opset 17

For more information on the download command, see the CLI Reference.