Denoising Pretraining for Semantic Segmentation

Citations Over TimeTop 10% of 2022 papers

Abstract

Semantic segmentation labels are expensive and time consuming to acquire. To improve label efficiency of semantic segmentation models, we revisit denoising autoencoders and study the use of a denoising objective for pretraining UNets. We pretrain a Transformer-based UNet as a denoising autoencoder, followed by fine-tuning on semantic segmentation using few labeled examples. Denoising pretraining outperforms training from random initialization, and even supervised ImageNet-21K pretraining of the encoder when the number of labeled images is small. A key advantage of denoising pretraining over supervised pretraining of the backbone is the ability to pretrain the decoder, which would otherwise be randomly initialized. We thus propose a novel Decoder Denoising Pretraining (DDeP) method, in which we initialize the encoder using supervised learning and pretrain only the decoder using the denoising objective. Despite its simplicity, DDeP achieves state-of-the-art results on label-efficient semantic segmentation, offering considerable gains on the Cityscapes, Pascal Context, and ADE20K datasets.

Related Papers

• → Entraînement auto-supervisé pour débruitage aveuble de video(2021)32 cited
• → Video denoising via online sparse and low-rank matrix decomposition(2016)14 cited
• → An ICI Based Algorithm for Fast Denoising of Video Signals(2014)11 cited
• Simulation on the Seeker Signal Wavelets Denoising Based on Parameters Optimization(2010)
• → Multiple forward-backward strategies for two-stage video denoising(2023)