TouchDesigner-YOLOv7

Prerequisite

• TouchDesinger's Release Notes

Build version : 2022.28040
TouchDesigner Python version: 3.9.5
TouchDesigner CUDA version: 11.2
TouchDesigner cuDNN version: 8.1.1

CUDA, cuDNN

1. Install CUDA Toolkit 11.2 Update 2 Downloads
2. Download cuDNN 8.1.1 for CUDA 11.2
3. Unzip cudnn-11.2-windows-x64-v8.1.1.33.zip
4. Copy files (cudnn-11.2-windows-x64-v8.1.1.33\cuda) to C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.2

Anaconda

1. Install Anaconda Distribution
2. Run “Anaconda Prompt” application
3. Create virtual environment conda create -n td python=3.9.5
4. Activate virtual environment conda activate td
5. Install onnxruntime-gpu pip install onnxruntime-gpu

YOLOv7

• Official Repository

TouchDesigner

• Add External Python to Search Path
  [Edit]→[Preferences]→[General]→[Add External Python to Search Path]
  Set C:/Users/USERNAME/anaconda3/envs/ENV_NAME/Lib/site-packages

• Code

  import numpy as np
  import onnxruntime as ort
  
  class YOLOv7:
  	def __init__(self, model):
  		providers = ['CUDAExecutionProvider', 'CPUExecutionProvider']
  		self.session = ort.InferenceSession(model, providers=providers)
  
  	def run(self):
  		image = op('src').numpyArray()
  		image = image[:,:,:3]
  		image = image.transpose((2,0,1))
  		image = np.expand_dims(image, 0)
  		image = np.ascontiguousarray(image)
  		
  		outputs = self.session.run(['output'], {'images':image})[0]
  
  		op('output').clear()
  		op('output').appendRows(outputs)
  		
  yolo = YOLOv7('./yolov7-tiny.onnx')
  yolo.run()