ANPR Recognition Errors on Jetson Orin Nano

Issue Overview

Users are encountering errors while attempting to run Automatic Number Plate Recognition (ANPR) on the Nvidia Jetson Orin Nano development board. The issue occurs when executing a specific Python command for license plate detection and recognition using the jetson-inference library. The problem appears during the setup or execution of the ANPR system, potentially impacting the functionality of the license plate recognition process.

Key details:

• Hardware: Jetson Orin Nano 8GB
• Software: JetPack 5.1.1, Ubuntu 20.04
• Command causing issues:

python3 detectnet-camera.py --model=./networks/az_ocr/az_ocr_ssdmobilenetv1_2.onnx --class_labels=./networks/az_ocr/labels.txt --input_blob=input_0 --output_cvg=scores --output_bbox=boxes --camera=/dev/video0 --width=640 --height=480

Possible Causes

1. Incompatible model or library versions: The ONNX model or the jetson-inference library may not be compatible with the specific Jetson Orin Nano hardware or JetPack version.

2. Missing dependencies: Required libraries or dependencies for running the ANPR system might not be installed or properly configured.

3. Camera configuration issues: The specified camera device (/dev/video0) might not be correctly recognized or accessible.

4. File path errors: The paths to the model file or labels file might be incorrect or the files might be missing.

5. JetPack or CUDA incompatibilities: The version of JetPack or CUDA installed might not be fully compatible with the ANPR software or the jetson-inference library.

6. Memory constraints: The 8GB RAM of the Jetson Orin Nano might be insufficient for running the ANPR model, causing out-of-memory errors.

7. Incorrect command parameters: The command-line arguments might be incorrectly specified or incompatible with the current setup.

Troubleshooting Steps, Solutions & Fixes

1. Verify hardware and software compatibility:

   • Confirm that the ANPR system and jetson-inference library support Jetson Orin Nano and JetPack 5.1.1.
   • Check the GitHub repository for any specific requirements or known issues with your hardware/software combination.

2. Update JetPack and CUDA:

   • Ensure you have the latest JetPack version compatible with your Jetson Orin Nano.
   • Update CUDA to the latest version supported by your JetPack:

     sudo apt update
     sudo apt upgrade
     

3. Install required dependencies:

   • Install any missing Python libraries:

     pip3 install -r requirements.txt
     

   • If the requirements.txt file is not available, check the GitHub repository for a list of required packages.

4. Verify camera setup:

   • Check if the camera is properly connected and recognized:

     ls /dev/video*
     

   • If /dev/video0 is not listed, try other video devices or check camera drivers.

5. Confirm file paths:

   • Verify that the model and labels files exist in the specified locations:

     ls ./networks/az_ocr/az_ocr_ssdmobilenetv1_2.onnx
     ls ./networks/az_ocr/labels.txt
     

   • If files are missing, download them from the GitHub repository.

6. Check system resources:

   • Monitor RAM usage during execution:

     free -h
     

   • If memory is insufficient, try closing other applications or reducing the model size.

7. Modify command parameters:

   • Try running the command with default parameters first, then gradually add specific options:

     python3 detectnet-camera.py --model=./networks/az_ocr/az_ocr_ssdmobilenetv1_2.onnx --class_labels=./networks/az_ocr/labels.txt
     

8. Capture complete error logs:

   • Run the command with added verbosity and redirect output to a file:

     python3 detectnet-camera.py --model=./networks/az_ocr/az_ocr_ssdmobilenetv1_2.onnx --class_labels=./networks/az_ocr/labels.txt --input_blob=input_0 --output_cvg=scores --output_bbox=boxes --camera=/dev/video0 --width=640 --height=480 --verbose 2>&1 | tee error_log.txt
     

   • Analyze the error_log.txt file for specific error messages and stack traces.

9. Check GitHub issues and documentation:

   • Search the GitHub repository’s issues section for similar problems and potential solutions.
   • Review the documentation for any specific setup instructions for Jetson Orin Nano.

10. Rebuild jetson-inference:

    • If possible, try rebuilding the jetson-inference library from source:

      cd jetson-inference
      mkdir build
      cd build
      cmake ../
      make -j$(nproc)
      sudo make install
      sudo ldconfig
      

If the issue persists after trying these steps, consider opening a new issue on the GitHub repository with detailed information about your setup, the complete error log, and the steps you’ve already taken to troubleshoot the problem.