[NeurIPS 2021] SSUL: Semantic Segmentation with Unknown Label for Exemplar-based Class-Incremental Learning

Related tags

Deep LearningSSUL
Overview

SSUL - Official Pytorch Implementation (NeurIPS 2021)

SSUL: Semantic Segmentation with Unknown Label for Exemplar-based Class-Incremental Learning
Sungmin Cha1,2*, Beomyoung Kim3*, YoungJoon Yoo2,3, Taesup Moon1
* Equal contribution

1 Department of Electrical and Computer Engineering, Seoul National University
2 NAVER AI Lab
3 Face, NAVER Clova

NeurIPS 2021

Paper

Abtract

This paper introduces a solid state-of-the-art baseline for a class-incremental semantic segmentation (CISS) problem. While the recent CISS algorithms utilize variants of the knowledge distillation (KD) technique to tackle the problem, they failed to fully address the critical challenges in CISS causing the catastrophic forgetting; the semantic drift of the background class and the multi-label prediction issue. To better address these challenges, we propose a new method, dubbed SSUL-M (Semantic Segmentation with Unknown Label with Memory), by carefully combining techniques tailored for semantic segmentation. Specifically, we claim three main contributions. (1) defining unknown classes within the background class to help to learn future classes (help plasticity), (2) freezing backbone network and past classifiers with binary cross-entropy loss and pseudo-labeling to overcome catastrophic forgetting (help stability), and (3) utilizing tiny exemplar memory for the first time in CISS to improve both plasticity and stability. The extensively conducted experiments show the effectiveness of our method, achieving significantly better performance than the recent state-of-the-art baselines on the standard benchmark datasets. Furthermore, we justify our contributions with thorough ablation analyses and discuss different natures of the CISS problem compared to the traditional class-incremental learning targeting classification.

Experimental Results (mIoU all)

Method VOC 10-1 (11 tasks) VOC 15-1 (6 tasks) VOC 5-3 (6 tasks) VOC 19-1 (2 tasks) VOC 15-5 (2 tasks) VOC 5-1 (16 tasks) VOC 2-1 (19 tasks)
MiB 12.65 29.29 46.71 69.15 70.08 10.03 9.88
PLOP 30.45 54.64 18.68 73.54 70.09 6.46 4.47
SSUL 59.25 67.61 56.89 75.44 71.22 48.65 38.32
SSUL-M 64.12 71.37 58.37 76.49 73.02 55.11 44.74
Method ADE 100-5 (11 tasks) ADE 100-10 (6 tasks) ADE 100-50 (2 tasks) ADE 50-50 (3 tasks)
MiB 25.96 29.24 32.79 29.31
PLOP 28.75 31.59 32.94 30.40
SSUL 32.48 33.10 33.58 29.56
SSUL-M 34.56 34.46 34.37 29.77

Getting Started

Requirements

  • torch>=1.7.1
  • torchvision>=0.8.2
  • numpy
  • pillow
  • scikit-learn
  • tqdm
  • matplotlib

Datasets

data_root/
    --- VOC2012/
        --- Annotations/
        --- ImageSet/
        --- JPEGImages/
        --- SegmentationClassAug/
        --- saliency_map/
    --- ADEChallengeData2016
        --- annotations
            --- training
            --- validation
        --- images
            --- training
            --- validation

Download SegmentationClassAug and saliency_map

Class-Incremental Segmentation Segmentation on VOC 2012

DATA_ROOT=your_dataset_root_path
DATASET=voc
TASK=15-1 # [15-1, 10-1, 19-1, 15-5, 5-3, 5-1, 2-1, 2-2]
EPOCH=50
BATCH=32
LOSS=bce_loss
LR=0.01
THRESH=0.7
MEMORY=100 # [0 (for SSUL), 100 (for SSUL-M)]

python main.py --data_root ${DATA_ROOT} --model deeplabv3_resnet101 --gpu_id 0,1 --crop_val --lr ${LR} --batch_size ${BATCH} --train_epoch ${EPOCH} --loss_type ${LOSS} --dataset ${DATASET} --task ${TASK} --overlap --lr_policy poly --pseudo --pseudo_thresh ${THRESH} --freeze --bn_freeze --unknown --w_transfer --amp --mem_size ${MEMORY}

Class-Incremental Segmentation Segmentation on ADE20K

DATA_ROOT=your_dataset_root_path
DATASET=ade
TASK=100-5 # [100-5, 100-10, 100-50, 50-50]
EPOCH=100
BATCH=24
LOSS=bce_loss
LR=0.05
THRESH=0.7
MEMORY=300 # [0 (for SSUL), 300 (for SSUL-M)]

python main.py --data_root ${DATA_ROOT} --model deeplabv3_resnet101 --gpu_id 0,1 --crop_val --lr ${LR} --batch_size ${BATCH} --train_epoch ${EPOCH} --loss_type ${LOSS} --dataset ${DATASET} --task ${TASK} --overlap --lr_policy warm_poly --pseudo --pseudo_thresh ${THRESH} --freeze --bn_freeze --unknown --w_transfer --amp --mem_size ${MEMORY}

Qualitative Results

Acknowledgement

Our implementation is based on these repositories: DeepLabV3Plus-Pytorch, Torchvision.

License

SSUL
Copyright 2021-present NAVER Corp.

Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in
all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
THE SOFTWARE.
Owner
Clova AI Research
Open source repository of Clova AI Research, NAVER & LINE
Clova AI Research
Distributed Arcface Training in Pytorch

Distributed Arcface Training in Pytorch

3 Nov 23, 2021
Framework to build and train RL algorithms

RayLink RayLink is a RL framework used to build and train RL algorithms. RayLink was used to build a RL framework, and tested in a large-scale multi-a

Bytedance Inc. 32 Oct 07, 2022
Towards End-to-end Video-based Eye Tracking

Towards End-to-end Video-based Eye Tracking The code accompanying our ECCV 2020 publication and dataset, EVE. Authors: Seonwook Park, Emre Aksan, Xuco

Seonwook Park 76 Dec 12, 2022
Public repository created to store my custom-made tools for Just Dance (UbiArt Engine)

Woody's Just Dance Tools Public repository created to store my custom-made tools for Just Dance (UbiArt Engine) Development and updates Almost all of

Wodson de Andrade 8 Dec 24, 2022
Semi-Supervised Semantic Segmentation via Adaptive Equalization Learning, NeurIPS 2021 (Spotlight)

Semi-Supervised Semantic Segmentation via Adaptive Equalization Learning, NeurIPS 2021 (Spotlight) Abstract Due to the limited and even imbalanced dat

Hanzhe Hu 99 Dec 12, 2022
3D Human Pose Machines with Self-supervised Learning

3D Human Pose Machines with Self-supervised Learning Keze Wang, Liang Lin, Chenhan Jiang, Chen Qian, and Pengxu Wei, “3D Human Pose Machines with Self

Chenhan Jiang 398 Dec 20, 2022
An implementation of a sequence to sequence neural network using an encoder-decoder

Keras implementation of a sequence to sequence model for time series prediction using an encoder-decoder architecture. I created this post to share a

Luke Tonin 195 Dec 17, 2022
KoCLIP: Korean port of OpenAI CLIP, in Flax

KoCLIP This repository contains code for KoCLIP, a Korean port of OpenAI's CLIP. This project was conducted as part of Hugging Face's Flax/JAX communi

Jake Tae 100 Jan 02, 2023
Unrolled Variational Bayesian Algorithm for Image Blind Deconvolution

unfoldedVBA Unrolled Variational Bayesian Algorithm for Image Blind Deconvolution This repository contains the Pytorch implementation of the unrolled

Yunshi HUANG 2 Jul 10, 2022
PyTorch Implementation of Unsupervised Depth Completion with Calibrated Backprojection Layers (ORAL, ICCV 2021)

Unsupervised Depth Completion with Calibrated Backprojection Layers PyTorch implementation of Unsupervised Depth Completion with Calibrated Backprojec

80 Dec 13, 2022
SimDeblur is a simple framework for image and video deblurring, implemented by PyTorch

SimDeblur (Simple Deblurring) is an open source framework for image and video deblurring toolbox based on PyTorch, which contains most deep-learning based state-of-the-art deblurring algorithms. It i

220 Jan 07, 2023
Official code for the CVPR 2022 (oral) paper "Extracting Triangular 3D Models, Materials, and Lighting From Images".

nvdiffrec Joint optimization of topology, materials and lighting from multi-view image observations as described in the paper Extracting Triangular 3D

NVIDIA Research Projects 1.4k Jan 01, 2023
Beta Shapley: a Unified and Noise-reduced Data Valuation Framework for Machine Learning

Beta Shapley: a Unified and Noise-reduced Data Valuation Framework for Machine Learning This repository provides an implementation of the paper Beta S

Yongchan Kwon 28 Nov 10, 2022
PyTorch Connectomics: segmentation toolbox for EM connectomics

Introduction The field of connectomics aims to reconstruct the wiring diagram of the brain by mapping the neural connections at the level of individua

Zudi Lin 132 Dec 26, 2022
a reccurrent neural netowrk that when trained on a peice of text and fed a starting prompt will write its on 250 character text using LSTM layers

RNN-Playwrite a reccurrent neural netowrk that when trained on a peice of text and fed a starting prompt will write its on 250 character text using LS

Arno Barton 1 Oct 29, 2021
Official implementation of the paper Image Generators with Conditionally-Independent Pixel Synthesis https://arxiv.org/abs/2011.13775

CIPS -- Official Pytorch Implementation of the paper Image Generators with Conditionally-Independent Pixel Synthesis Requirements pip install -r requi

Multimodal Lab @ Samsung AI Center Moscow 201 Dec 21, 2022
A Learning-based Camera Calibration Toolbox

Learning-based Camera Calibration A Learning-based Camera Calibration Toolbox Paper The pdf file can be found here. @misc{zhang2022learningbased,

Eason 14 Dec 21, 2022
PyTorch implementation of federated learning framework based on the acceleration of global momentum

Federated Learning with Acceleration of Global Momentum PyTorch implementation of federated learning framework based on the acceleration of global mom

0 Dec 23, 2021
Activity image-based video retrieval

Cross-modal-retrieval Our approach is focus on Activity Image-to-Video Retrieval (AIVR) task. The compared methods are state-of-the-art single modalit

BCMI 75 Oct 21, 2021
Efficient semidefinite bounds for multi-label discrete graphical models.

Low rank solvers #################################### benchmark/ : folder with the random instances used in the paper. ############################

1 Dec 08, 2022