EventDenoising

Official implementation of "On Spatiotemporal Relation Modeling of Event: Parallel Denoising and No-Reference Evaluation".

🚀 Quick Start

Environment

Python 3.10
Pytorch 2.2.1
CUDA 11.8
cudnn 8

Installation

conda create -n EventDenoising python=3.10
conda activate EventDenoising
pip install -r requirements.txt

Require additional install mamba-ssm==2.2.2 and pytorch3d==0.7.5.

EDformer++ Evaluation

# Testing SNR with DVSCLEAN Dataset
python eval_DVSCLEAN.py

# Testing AUC with DND21 Dataset
python eval_DND21.py

ENRR Evaluation

# Testing ENR with DND21 Dataset
python eval_enrr_DND21.py

# Testing ENR with ED24 Dataset
python eval_enrr_ED24.py

🏋️ Training

# EDfromer++ train
python train_edformer_plus.py

# ENRR train
python train_enrr.py

📦 Datasets

You can use the sample data in ./data for quick testing, or download the complete dataset from the link below for full evaluation or model training.

Dataset	Usage
ED24	EDformer++ and ENRR Training
DND21	AUC and ENR Evaluation
DVSCLEAN	SNR Evaluation
ED-KoGTL	SNR Evaluation
E-MLB	MESR Evaluation

⁉️ Contact

For technical issues, please contact: 📧 jiangbin@smail.nju.edu.cn