Training of Corner Detection model

This Python script allows you to train a corner detection model from a set of RGB images and an event simulator.

Expected Output

Corner detection training results:

  • checkpoints (models at different training stages)

  • log files

  • video demos on test dataset

Setup & requirements

To run the script, you need:

  • path to the output directory

  • path to the training dataset:

    • a folder containing 2 sub folders, named train and val. Each subfolder should contain multiple .jpg or .png files.

How to start

To run the script:

Linux

python3 train_corner_detection.py --root_dir /path/to/output --dataset_path /path/to/dataset

Windows

python train_corner_detection.py --root_dir /path/to/output --dataset_path /path/to/dataset

Note

Adding an optional argument --randomize_noises to randomize the noise in the simulated data is recommended.

To find the full list of options, run:

Linux

python3 train_corner_detection.py -h

Windows

python train_corner_detection.py -h

Warning

In case of “CUDA out of memory” error, you can try to reduce the batch-size, num-tbins, height or width parameters.