Skip to content

Latest commit

 

History

History
116 lines (76 loc) · 1.92 KB

YOLOR.md

File metadata and controls

116 lines (76 loc) · 1.92 KB
Raw

YOLOR usage

Convert model

1. Download the YOLOR repo and install the requirements

git clone https://github.com/WongKinYiu/yolor.git
cd yolor
pip3 install -r requirements.txt
pip3 install onnx onnxsim onnxruntime

NOTE: It is recommended to use Python virtualenv.

2. Copy conversor

Copy the export_yolor.py file from models-convertion/yolo/onnx directory to the yolor folder.

3. Download the model

Download the pt file from YOLOR repo.

NOTE: You can use instead your trained model weights.

4. Convert model

Generate the ONNX model file

  • Main branch

    Example for YOLOR-CSP

    python3 export_yolor.py -w yolor_csp.pt -c cfg/yolor_csp.cfg --dynamic

  • Paper branch

    Example for YOLOR-P6

    python3 export_yolor.py -w yolor-p6.pt --dynamic

NOTE: To convert a P6 model

--p6

NOTE: To change the inference size (defaut: 640 / 1280 for --p6 models)

-s SIZE
--size SIZE
-s HEIGHT WIDTH
--size HEIGHT WIDTH

Example for 1280

-s 1280

or

-s 1280 1280

NOTE: To simplify the ONNX model

--simplify

NOTE: To use dynamic batch-size

--dynamic

NOTE: To use implicit batch-size (example for batch-size = 4)

--batch 4

NOTE: The default opset is 12.

--opset 12

5. Upload the generated ONNX file

On Lumeo's Console click Design -> AI Models -> Add model and fill the form using those settings:

Types & Weights tab

  • Format: ONNX Lumeo YOLO
  • Capability: Detection
  • Architecture: YOLOR
  • Labels: Type the labels or import the labels file (each line represents a different model's class)
  • Weights: Upload the previously generated ONNX file

Parameters tab

  • Net scale factor: 0.0039215698 (click in 1/256)
  • Color format: RGB
  • Network precision: Float16
  • Clustering algorithm: NMS