Visualization of Preprocessed HDF5 Tensor Files

This Python script allows you to visualize HDF5 tensor files, containing preprocessed event tensors. If bounding boxes are provided, they will be visualized altogether in the detection mode.

The source code of this sample can be found in <install-prefix>/share/metavision/sdk/ml/python_samples/viz_data when installing Metavision SDK from installer or packages. For other deployment methods, check the page Path of Samples.

Expected Output

A video showing the preprocessed event data in HDF5 tensor format (h5).

Setup & requirements

To run the script, you need to provide the path to a folder containing a label_map_dictionary.json file and subfolders named train, val or test where one or multiple h5 files are located.

Those HDF5 tensor files can be created with our generate_hdf5 script as explained in the tutorial Precomputing features as HDF5 tensor files.

Note

With --video-output defined, a video will be saved locally.

How to start

To run the script:

python viz_data.py /path/to/folder_with_hdf5_files

To find the full list of options, run:

python viz_data.py -h