VDSR
VDSR is a milestone architecture in the development history of image super-resolution technology. It successfully increased the depth of convolutional neural networks to 20 layers for the first time, solving the convergence problem of deep networks by introducing global residual learning mechanisms, and establishing the core approach of learning "image residuals" rather than directly learning pixel values.
- Key Features: Supports multi-scale (2x, 3x, 4x) image super-resolution reconstruction with a single model, possessing powerful edge recovery capabilities and texture detail enhancement effects. Widely used in high-definition video conversion, digital image restoration, and medical image enhancement.
- Version Note: This case uses the standard VDSR architecture. The model effectively captures image context information through a large receptive field and achieves efficient training processes using high learning rates with gradient clipping techniques. As the pioneering work of deep super-resolution algorithms, it provides visual clarity far beyond traditional interpolation methods while maintaining structural simplicity, making it the classic cornerstone choice for studying super-resolution technology evolution and industrial deployment.
Environment Setup
Configure the required environment in advance.
Quick Start
Download Model Files
O6 / O6N
cd ai_model_hub_25_Q3/models/ComputeVision/Super_Resolution/onnx_vdsr
wget https://www.modelscope.cn/models/cix/ai_model_hub_25_Q3/resolve/master/models/ComputeVision/Super_Resolution/onnx_vdsr/vdsr.cix
Model Testing
info
Activate the virtual environment before running!
O6 / O6N
python3 inference_npu.py
Complete Conversion Workflow
Download Model Files
Linux PC
cd ai_model_hub_25_Q3/models/ComputeVision/Super_Resolution/onnx_vdsr/model
wget https://www.modelscope.cn/models/cix/ai_model_hub_25_Q3/resolve/master/models/ComputeVision/Super_Resolution/onnx_vdsr/model/vdsr.onnx
Project Structure
├── cfg
├── datasets
├── inference_npu.py
├── inference_onnx.py
├── model
├── ReadMe.md
├── test_data
└── vdsr.cix
Perform Model Quantization and Conversion
Linux PC
cd ..
cixbuild cfg/onnx_vdsr_build.cfg
Push to Board
After completing the model conversion, push the cix model file to the board.
Test Host Inference
Run Inference Script
Linux PC
python3 inference_onnx.py
Model Inference Results
Deploy to NPU
Run Inference Script
O6 / O6N
python3 inference_npu.py
Model Inference Results
O6 / O6N
$ python3 inference_npu.py
npu: noe_init_context success
npu: noe_load_graph success
Input tensor count is 1.
Output tensor count is 1.
npu: noe_create_job success
npu: noe_clean_job success
npu: noe_unload_graph success
npu: noe_deinit_context success
