Deep Active Learning for Joint Classification & Segmentation with Weak Annotator

@article{Belharbi2021DeepAL,
  title={Deep Active Learning for Joint Classification \& Segmentation with Weak Annotator},
  author={Soufiane Belharbi and Ismail Ben Ayed and Luke McCaffrey and {\'E}ric Granger},
  journal={2021 IEEE Winter Conference on Applications of Computer Vision (WACV)},
  year={2021},
  pages={3337-3346}
}
CNN visualization and interpretation methods, like class-activation maps (CAMs), are typically used to highlight the image regions linked to class predictions. These models allow to simultaneously classify images and extract class-dependent saliency maps, without the need for costly pixel-level annotations. However, they typically yield segmentations with high false-positive rates and, therefore, coarse visualisations, more so when processing challenging images, as encountered in histology. To… 
To Actively Initialize Active Learning
Deep Weakly-Supervised Learning Methods for Classification and Localization in Histology Images: A Comparative Study
TLDR
Four key challenges are identi-fied for the application of deep WSOL methods in histology – under/over activation of CAMs, sensitivity to thresholding, and model selection – and research avenues are provided to mitigate these issues.
WSSS4LUAD: Grand Challenge on Weakly-supervised Tissue Semantic Segmentation for Lung Adenocarcinoma
TLDR
WSSS approaches with patch-level annotations can be a complement to the traditional pixel annotations while reducing the annotation efforts, according to the technical reports of the top-tier teams, and CAM is still the most popular approach in WSSS.
Negative Evidence Matters in Interpretable Histology Image Classification
of using both positive and negative evidence, the first obtained from a classifier, and the other naturally available in datasets.
Multi-Layer Pseudo-Supervision for Histopathology Tissue Semantic Segmentation using Patch-level Classification Labels
F-CAM: Full Resolution Class Activation Maps via Guided Parametric Upscaling
TLDR
This work proposes a trainable decoding architecture that can be connected to any CNN classifier to produce highly accurate CAM localizations, and introduces a generic method for parametric upscaling of CAMs that allows constructing accurate full resolution CAMs (FCAMs).
Deep Interpretable Classification and Weakly-Supervised Segmentation of Histology Images via Max-Min Uncertainty
TLDR
High uncertainty is introduced as a criterion to localize non-discriminative regions that do not affect classifier decision, and is described with original Kullback-Leibler (KL) divergence losses evaluating the deviation of posterior predictions from the uniform distribution.
Food Region Extraction Based on Saliency Detection Model
TLDR
A method that can automatically extract food regions from food images by using the saliency detection model based on a deep neural network (DNN) and the Saliency thresholding method based on the average saliency value is proposed.
Practical Active Learning with Model Selection for Small Data
TLDR
This work focuses on the case of having a very small labeling budget, and develops a simple and fast method for practical active learning with model selection, based on an underlying pool-based active learner for binary classification using support vector classification with a radial basis function kernel.
CoNeRF: Controllable Neural Radiance Fields
TLDR
The key idea is to treat the attributes as latent variables that are regressed by the neural network given the scene encoding, which leads to a few-shot learning framework, where attributes are discovered automatically by the framework, when annotations are not provided.
...
...

References

SHOWING 1-10 OF 88 REFERENCES
Deep weakly-supervised learning methods for classification and localization in histology images: a survey
TLDR
Results indicate that several deep learning models, and in particular WILDCAT and deep MIL can provide a high level of classification accuracy, although pixel-wise localization of cancer regions remains an issue for such images.
Suggestive Annotation: A Deep Active Learning Framework for Biomedical Image Segmentation
TLDR
A deep active learning framework that combines fully convolutional network (FCN) and active learning to significantly reduce annotation effort by making judicious suggestions on the most effective annotation areas is presented.
Active deep learning for medical imaging segmentation
This thesis proposes a novel active learning framework capable to train effectively a convolutional neural network for semantic segmentation of medical imaging, with a limited amount of training
From image-level to pixel-level labeling with Convolutional Networks
TLDR
A Convolutional Neural Network-based model is proposed, which is constrained during training to put more weight on pixels which are important for classifying the image, and which beats the state of the art results in weakly supervised object segmentation task by a large margin.
Cost-Effective Active Learning for Deep Image Classification
TLDR
This paper proposes a novel active learning (AL) framework, which is capable of building a competitive classifier with optimal feature representation via a limited amount of labeled training instances in an incremental learning manner and incorporates deep convolutional neural networks into AL.
Constrained Deep Weak Supervision for Histopathology Image Segmentation
TLDR
A new weakly supervised learning algorithm to learn to segment cancerous regions in histopathology images with fully convolutional networks and an effective way to introduce constraints to the authors' neural networks to assist the learning process is proposed.
Min-max Entropy for Weakly Supervised Pointwise Localization
TLDR
A new deep learning method is proposed for WSL, composed of a localizer and a classifier, where the localizer is constrained to determine relevant and irrelevant regions using conditional entropy (CE) with the aim to reduce false positive regions.
Deep Active Learning for Axon-Myelin Segmentation on Histology Data
TLDR
A framework for Deep Active Learning applied to a real-world scenario that relies on the U-Net architecture and overall uncertainty measure to suggest which sample to annotate, and shows evidence that this straightforward implementation achieves a high segmentation performance with very few labelled samples.
Normalized Cut Loss for Weakly-Supervised CNN Segmentation
TLDR
This work proposes a new principled loss function evaluating network output with criteria standard in "shallow" segmentation, e.g. normalized cut which evaluates only seeds where labels are known while normalized cut softly evaluates consistency of all pixels.
The Power of Ensembles for Active Learning in Image Classification
TLDR
It is found that ensembles perform better and lead to more calibrated predictive uncertainties, which are the basis for many active learning algorithms, and Monte-Carlo Dropout uncertainties perform worse.
...
...