1.0.8

What's Changed

• add enable_profiling by @fateshelled in #3

New Contributors

• @fateshelled made their first contribution in #3

Full Changelog: 1.0.7...1.0.8

1.0.7

• Increased supported data types

  ONNX_DTYPES_TO_NUMPY_DTYPES: dict = {
      f'{onnx.TensorProto.FLOAT16}': np.float16,
      f'{onnx.TensorProto.FLOAT}': np.float32,
      f'{onnx.TensorProto.DOUBLE}': np.float64,
      f'{onnx.TensorProto.UINT8}': np.uint8,
      f'{onnx.TensorProto.UINT16}': np.uint16,
      f'{onnx.TensorProto.UINT32}': np.uint32,
      f'{onnx.TensorProto.UINT64}': np.uint64,
      f'{onnx.TensorProto.INT8}': np.int8,
      f'{onnx.TensorProto.INT16}': np.int16,
      f'{onnx.TensorProto.INT32}': np.int32,
      f'{onnx.TensorProto.INT64}': np.int64,
  }

What's Changed

• Increased supported data types by @PINTO0309 in #2

Full Changelog: 1.0.6...1.0.7

1.0.6

• Added intra_op_num_threads option

    -iont, --intra_op_num_threads INTRA_OP_NUM_THREADS
        Sets the number of threads used to parallelize the execution within nodes.
        Default is 0 to let onnxruntime choose.
  

Full Changelog: 1.0.5...1.0.6

1.0.5

• Support for SequenceConstruct

  INFO: file: yolopv2_192x320_test.onnx
  INFO: providers: ['CUDAExecutionProvider', 'CPUExecutionProvider']
  INFO: input_name.1: input shape: [1, 3, 192, 320] dtype: float32
  INFO: test_loop_count: 10
  INFO: total elapsed time:  76.85065269470215 ms
  INFO: avg elapsed time per pred:  7.685065269470215 ms
  INFO: output_name.1-0: pred-0 shape: [1, 255, 24, 40] dtype: float32
  INFO: output_name.1-1: pred-1 shape: [1, 255, 12, 20] dtype: float32
  INFO: output_name.1-2: pred-2 shape: [1, 255, 6, 10] dtype: float32
  INFO: output_name.2: anchor_grid0 shape: [1, 3, 1, 1, 2] dtype: float32
  INFO: output_name.3: anchor_grid1 shape: [1, 3, 1, 1, 2] dtype: float32
  INFO: output_name.4: anchor_grid2 shape: [1, 3, 1, 1, 2] dtype: float32
  INFO: output_name.5: seg shape: [1, 2, 192, 320] dtype: float32
  INFO: output_name.6: ll shape: [1, 1, 192, 320] dtype: float32
  

  
• Add short form

  usage:
    sit4onnx [-h]
    -if INPUT_ONNX_FILE_PATH
    [-b BATCH_SIZE]
    [-fs DIM0 [DIM1 DIM2 ...]]
    [-tlc TEST_LOOP_COUNT]
    [-oep {tensorrt,cuda,openvino_cpu,openvino_gpu,cpu}]
    [-ifp INPUT_NUMPY_FILE_PATHS_FOR_TESTING]
    [-ofp]
    [-n]
  

1.0.4

• Security update
• README update

1.0.3

• Allow static fixed shapes to be specified when dimensions other than batch size are undefined.
• Bug fixes.

1.0.2

Add an interface to allow arbitrary test data to be specified as input parameters.

1. numpy.ndarray
2. numpy file

1.0.1

• Returns numpy.ndarray of the last inference result as a return value when called from a Python script.
• Add --output_numpy_file option. Output the final inference results to a numpy file.
• Add --non_verbose option.

1.0.0

• Initial release