Adaptive feature recombination and recalibration for semantic segmentation with Fully Convolutional Networks

Fully Convolutional Networks have been achieving remarkable results in image semantic segmentation, while being efficient. Such efficiency results from the capability of segmenting several voxels in a single forward pass. So, there is a direct spatial correspondence between a unit in a feature map and the voxel in the same location. In a convolutional layer, the kernel spans over all channels and extracts information from them. We observe that linear recombination of feature maps by increasing the number of channels followed by compression may enhance their discriminative power. Moreover, not all feature maps have the same relevance for the classes being predicted. In order to learn the inter-channel relationships and recalibrate the channels to suppress the less relevant ones, Squeeze and Excitation blocks were proposed in the context of image classification with Convolutional Neural Networks. However, this is not well adapted for segmentation with Fully Convolutional Networks since they segment several objects simultaneously, hence a feature map may contain relevant information only in some locations. In this paper, we propose recombination of features and a spatially adaptive recalibration block that is adapted for semantic segmentation with Fully Convolutional Networks - the SegSE block. Feature maps are recalibrated by considering the cross-channel information together with spatial relevance. Experimental results indicate that Recombination and Recalibration improve the results of a competitive baseline, and generalize across three different problems: brain tumor segmentation, stroke penumbra estimation, and ischemic stroke lesion outcome prediction. The obtained results are competitive or outperform the state of the art in the three applications.

Deep Dense and Convolutional Autoencoders for Unsupervised Anomaly Detection in Machine Condition Sounds

This technical report describes two methods that were developed for Task 2 of the DCASE 2020 challenge. The challenge involves an unsupervised learning to detect anomalous sounds, thus only normal machine working condition samples are available during the training process. The two methods involve deep autoencoders, based on dense and convolutional architectures that use melspectogram processed sound features. Experiments were held, using the six machine type datasets of the challenge. Overall, competitive results were achieved by the proposed dense and convolutional AE, outperforming the baseline challenge method.