UFLDv2
Ultra Fast Lane Detection (UFLD) is a class of ultra-fast deep learning algorithms focused on lane detection. It changes the traditional pixel-level segmentation approach by introducing an innovative row-based selection classification mechanism, transforming the detection task into a simple classification problem, greatly improving model runtime speed.
- Key Features: Focuses on real-time lane detection in road scenarios, capable of quickly and accurately outlining lane boundaries, providing core visual support for autonomous driving systems' lane keeping assistance (LKA) and lane departure warning (LDW).
- Version Note: This case uses the Ultra Fast Lane Detection V2 (UFLDv2) model. As an advanced version of this series, it introduces a hybrid anchor mechanism that not only enhances detection robustness for curves and complex occlusion scenarios but also maintains the series' consistent ultra-fast inference advantages. It further improves spatial structure capture capabilities while ensuring low latency, making it the current mainstream balanced choice for efficient real-time lane perception on in-vehicle embedded endpoints.
Environment Setup
Configure the required environment in advance.
Quick Start
Download Model Files
O6 / O6N
cd ai_model_hub_25_Q3/models/ComputeVision/Lane_Detection/onnx_Ultra_Fast_Lane_Detection_v2
wget https://www.modelscope.cn/models/cix/ai_model_hub_25_Q3/resolve/master/models/ComputeVision/Lane_Detection/onnx_Ultra_Fast_Lane_Detection_v2/Ultra_Fast_Lane_Detection_v2.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/Lane_Detection/onnx_Ultra_Fast_Lane_Detection_v2/model
wget https://www.modelscope.cn/models/cix/ai_model_hub_25_Q3/resolve/master/models/ComputeVision/Lane_Detection/onnx_Ultra_Fast_Lane_Detection_v2/model/Ultra_Fast_Lane_Detection_v2.onnx
Project Structure
├── cfg
├── datasets
├── inference_npu.py
├── inference_onnx.py
├── model
├── ReadMe.md
├── test_data
└── Ultra_Fast_Lane_Detection_v2.cix
Perform Model Quantization and Conversion
Linux PC
cd ..
cixbuild cfg/Ultra-Fast-Lane-Detection_v2build.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 4.
npu: noe_create_job success
npu: noe_clean_job success
npu: noe_unload_graph success
npu: noe_deinit_context success
