Dual-Level Adaptive Information Filtering for Interactive Image Segmentation

Ervine Zheng, Qi Yu, Rui Li, Pengcheng Shi, Anne Haake
Proceedings of The 25th International Conference on Artificial Intelligence and Statistics, PMLR 151:6862-6879, 2022.

Abstract

Image segmentation can be performed interactively by accepting user annotations to refine the segmentation. It seeks frequent feedback from humans, and the model is updated with a smaller batch of data in each iteration of the feedback loop. Such a training paradigm requires effective information filtering to guide the model so that it can encode vital information and avoid overfitting due to limited data and inherent heterogeneity and noises thereof. We propose an adaptive interactive segmentation framework to support user interaction while introducing dual-level information filtering to train a robust model. The framework integrates an encoder-decoder architecture with a style-aware augmentation module that applies augmentation to feature maps and customizes the segmentation prediction for different latent styles. It also applies a systematic label softening strategy to generate uncertainty-aware soft labels for model updates. Experiments on both medical and natural image segmentation tasks demonstrate the effectiveness of the proposed framework.

Cite this Paper

BibTeX
@InProceedings{pmlr-v151-zheng22b, title = { Dual-Level Adaptive Information Filtering for Interactive Image Segmentation }, author = {Zheng, Ervine and Yu, Qi and Li, Rui and Shi, Pengcheng and Haake, Anne}, booktitle = {Proceedings of The 25th International Conference on Artificial Intelligence and Statistics}, pages = {6862--6879}, year = {2022}, editor = {Camps-Valls, Gustau and Ruiz, Francisco J. R. and Valera, Isabel}, volume = {151}, series = {Proceedings of Machine Learning Research}, month = {28--30 Mar}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v151/zheng22b/zheng22b.pdf}, url = {https://proceedings.mlr.press/v151/zheng22b.html}, abstract = { Image segmentation can be performed interactively by accepting user annotations to refine the segmentation. It seeks frequent feedback from humans, and the model is updated with a smaller batch of data in each iteration of the feedback loop. Such a training paradigm requires effective information filtering to guide the model so that it can encode vital information and avoid overfitting due to limited data and inherent heterogeneity and noises thereof. We propose an adaptive interactive segmentation framework to support user interaction while introducing dual-level information filtering to train a robust model. The framework integrates an encoder-decoder architecture with a style-aware augmentation module that applies augmentation to feature maps and customizes the segmentation prediction for different latent styles. It also applies a systematic label softening strategy to generate uncertainty-aware soft labels for model updates. Experiments on both medical and natural image segmentation tasks demonstrate the effectiveness of the proposed framework. } }
Endnote
%0 Conference Paper %T Dual-Level Adaptive Information Filtering for Interactive Image Segmentation %A Ervine Zheng %A Qi Yu %A Rui Li %A Pengcheng Shi %A Anne Haake %B Proceedings of The 25th International Conference on Artificial Intelligence and Statistics %C Proceedings of Machine Learning Research %D 2022 %E Gustau Camps-Valls %E Francisco J. R. Ruiz %E Isabel Valera %F pmlr-v151-zheng22b %I PMLR %P 6862--6879 %U https://proceedings.mlr.press/v151/zheng22b.html %V 151 %X Image segmentation can be performed interactively by accepting user annotations to refine the segmentation. It seeks frequent feedback from humans, and the model is updated with a smaller batch of data in each iteration of the feedback loop. Such a training paradigm requires effective information filtering to guide the model so that it can encode vital information and avoid overfitting due to limited data and inherent heterogeneity and noises thereof. We propose an adaptive interactive segmentation framework to support user interaction while introducing dual-level information filtering to train a robust model. The framework integrates an encoder-decoder architecture with a style-aware augmentation module that applies augmentation to feature maps and customizes the segmentation prediction for different latent styles. It also applies a systematic label softening strategy to generate uncertainty-aware soft labels for model updates. Experiments on both medical and natural image segmentation tasks demonstrate the effectiveness of the proposed framework.
APA
Zheng, E., Yu, Q., Li, R., Shi, P. & Haake, A.. (2022). Dual-Level Adaptive Information Filtering for Interactive Image Segmentation . Proceedings of The 25th International Conference on Artificial Intelligence and Statistics, in Proceedings of Machine Learning Research 151:6862-6879 Available from https://proceedings.mlr.press/v151/zheng22b.html.

Related Material