Skip to content

Latest commit

 

History

History
61 lines (42 loc) · 2.13 KB

README.md

File metadata and controls

61 lines (42 loc) · 2.13 KB
Raw

Dataset Preparation

To prepare the datasets required for our hierarchical action segmentation benchmarks, follow the steps below. Each dataset should be placed in its corresponding directory under a main data directory.

Create Directory Structure

First, create the necessary directories:

mkdir -p data/Shot7M2
mkdir -p data/hBABEL
mkdir -p data/MABe22

Shot7M2

Download the Shot7M2 dataset from HuggingFace and place it in the data/Shot7M2 directory.

Steps:

  1. Ensure Git LFS is installed: 
    git lfs install
  2. Clone the dataset: 
    git clone https://huggingface.co/datasets/amathislab/SHOT7M2 data/Shot7M2

hBABEL Action Segmentation Benchmark

The hBABEL dataset is an extension of the BABEL dataset developed for hierarchical action segmentation. Please cite both the original BABEL paper and our ECCV paper when using hBABEL.

Steps:

A. Downloading Data

  1. Follow the instructions on the TEACH GitHub repository to download and process the AMASS and BABEL datasets.

  2. You should end up with the same data folders, as described in the TEACH repository.

B. Data Conversion to hBABEL

  1. Perform Procrustes alignment on the pose data: 
    python -m datasets.hBABEL.prepare_hbabel_data
  2. Generate the hBABEL dataset: 
    python datasets/hBABEL/prepare_hbabel_labels.py

MABe22

The MABe 2022 dataset is publicly available. Follow the steps below to download the mouse triplet data:

wget "https://data.caltech.edu/records/8kdn3-95j37/files/mouse_triplet_train.npy" -O "data/MABe22/mouse_triplet_train.npy"
wget "https://data.caltech.edu/records/8kdn3-95j37/files/mouse_triplet_test.npy" -O "data/MABe22/mouse_triplet_test.npy"
wget "https://data.caltech.edu/records/8kdn3-95j37/files/mouse_triplets_test_labels.npy" -O "data/MABe22/mouse_triplets_test_labels.npy"