Skip to content

Latest commit

 

History

History
47 lines (35 loc) · 1.47 KB

README.md

File metadata and controls

47 lines (35 loc) · 1.47 KB
Raw

ALIFE

This is the implementation of the paper "ALIFE: Adaptive Logit Regularizer and Feature Replay for Incremental Semantic Segmentation".

For detailed information, please check out our paper [arXiv].

Requirements

Getting started

git clone https://github.com/cvlab-yonsei/ALIFE.git
cd ALIFE

mkdir checkpoints logs
mkdir -p datasets/voc/19-1-ov datasets/voc/15-5-ov datasets/voc/15-5s-ov
mkdir -p datasets/ade/100-50-ov datasets/ade/50-ov datasets/ade/100-10-ov

Datasets

The structure should be organized as follows:

├─ ALIFE
└─ data
    ├─ ADEChallengeData2016
    └─ VOCdevkit

Training

Example commands

bash scripts/voc/alife.sh   # RUN ALIFE on PASCAL VOC
bash scripts/ade/alife-m.sh # RUN ALIFE-M on ADE20K

NOTE: we also provide individual scripts for each step of ALIFE (e.g., base stage, step 1, step 2, and step 3). You may need to modify those scripts or config files (Please see ./scripts/ or ./configs/) for running a specific scenario.

Acknowledgements

Our codes are partly based on the following repositories.