A library for easily integrating the EPIC-KITCHENS egocentric dataset in your experiments
$ pip install epic-kitchens- Segmentation scripts for splitting raw video frames/flow into action segments
- GulpIO adapter for ingesting and reading the dataset. This works particularly well for PyTorch and Tensorflow models.
- Dataset classes for loading and augmenting data
- Utilities for converting between frame indices for RGB and flow.
See the CHANGELOG for release notes
WARNING: If you have used v1.4.0 or below to gulp flow, please delete the data and regulp due to an issue in reading files from disk, the gulped frames will be unordered instead of ordered by frame index (i.e. lexicographically).
Check out the documentation on Read the docs