Awesome

[ECCV 2024] Versatile Incremental Learning: Towards Class and Domain-Agnostic Incremental Learning

Official PyTorch implementation for ECCV 2024 paper:

Versatile Incremental Learning: Towards Class and Domain-Agnostic Incremental Learning
Min-Yeong Park*, Jaeho Lee*, and Gyeong-Moon Park†

Environment

• Python 3.8.x
• PyTorch 1.12.1
• Torchvision 0.13.1
• NVIDIA GeForce RTX 3090
• CUDA 11.3

Getting Started

Environment

git clone git@github.com/KHU-AGI/VIL.git
cd VIL
conda create -n VIL python==3.8
conda activate VIL
pip install torch==1.12.1+cu113 torchvision==0.13.1+cu113 torchaudio==0.12.1 --extra-index-url https://download.pytorch.org/whl/cu113
pip install -r requirements.txt

Run ICON on VIL with iDigits dataset

python main.py --dataset iDigits --num_tasks 5 --seed 42 --versatile_inc --batch-size 24 --IC --thre 0.0 --beta 0.01 --use_cast_loss --k 2 --d_threshold