• State Resolved
  • Locked Locked
  • Replies 4 replies
  • Subscribers 99 subscribers
  • Views 168 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

TDA4AEN-Q1: DeepLab ONNX Model Compilation Issue

zhang xl
Prodigy 50 points

Part Number: TDA4AEN-Q1

I encountered an issue where quantization failed during the compilation of the DeepLab ONNX model. When the process reached the quantization stage, the log indicated that the compilation was completed successfully, but I could not find the output files in the corresponding folder.

  I adopt OSRT for compilation, and the corresponding configuration parameters are as follows.

"deeplabv3plus_mobilenetv2_tv_edgeailite_onnxrt_AM67A": create_model_config(
        task_type="segmentation",
        source=dict(),
        preprocess=dict(
            resize=512,
            crop=512,
            data_layout="NCHW",
            pad_color=0,
            resize_with_pad=False,
            reverse_channels=False,
        ),
        session=dict(
            session_name="onnxrt",
            model_path=os.path.join(
                models_base_path,
                "deeplab.onnx",
            ),
            meta_arch_type=3,
            input_mean=[123.675, 116.28, 103.53],
            input_scale=[0.017125, 0.017507, 0.017429],
            input_optimization=True,
        ),
        postprocess=dict(with_argmax=True),
        extra_info=dict(num_images=1, num_classes=2),
        runtime_options={ 
            'tensor_bits': 16,  
            'accuracy_level': 1,  
            'debug_level': 0,
            'advanced_options:calibration_frames': 2,  
            'advanced_options:calibration_iterations': 2,  
        }
    ),

image.png

  • 0
    TI__Prodigy 630 points

    Hi,

    Can you please confirm which branch/tag of edgeai-tidl-tools you are using?

  • 0
    Prodigy 50 points

    The version I am using is re11.01.

  • 0 in reply to Abhay Chirania
    Prodigy 50 points

    Hello, the unquantifiable compilation issue has been resolved. It was caused by incorrect settings for the number of images used in quantitative compilation. However, I have encountered a new problem. When performing quantitative compilation with the ONNX model downloaded from TI's official GitHub repository, the following error occurred. I have successfully completed quantitative compilation with this same ONNX model before, but now I keep getting errors indicating that a certain operator is not supported. I have uploaded the ONNX file I used and the parameters for quantitative compilation for your reference. Could you please help check what caused this problem?

    Fullscreen 4188.config.txt Download 
    "deeplabv3plus_mobilenetv2_tv_edgeailite_onnxrt_AM67A": create_model_config(
        task_type="segmentation",
        source=dict(),
        preprocess=dict(
            resize=512,
            crop=512,
            data_layout="NCHW",
            pad_color=0,
            resize_with_pad=False,
            reverse_channels=False,
        ),
        session=dict(
            session_name="onnxrt",
            model_path=os.path.join(
                models_base_path,
                "deeplab.onnx",
            ),
            meta_arch_type=3,
            input_mean=[123.675, 116.28, 103.53],
            input_scale=[0.017125, 0.017507, 0.017429],
            input_optimization=True,
        ),
        postprocess=dict(with_argmax=True),
        extra_info=dict(num_images=2, num_classes=2),
        runtime_options={ 
            'tensor_bits': 16,  
            'accuracy_level': 1,  
            'debug_level': 0,
            'advanced_options:calibration_frames': 2,  
            'advanced_options:calibration_iterations': 2,  
        }
    ),
    deeplabv3plus_mobilenetv2_edgeailite_512x512_20210405.rar