By learning view synthesis, we explicitly encourage the feature extractor to encode information about not only the visible, but also the occluded parts of the scene.
Without the need of anomalous training images, we propose Convolutional Adversarial Variational autoencoder with Guided Attention (CAVGA), which localizes the anomaly with a convolutional latent variable to preserve the spatial information.
Ranked #18 on Anomaly Detection on MVTec AD (Segmentation AUROC metric, using extra training data)
Incremental learning (IL) is an important task aimed at increasing the capability of a trained model, in terms of the number of classes recognizable by the model.
Recent developments in gradient-based attention modeling have seen attention maps emerge as a powerful tool for interpreting convolutional neural networks.
Weakly supervised learning with only coarse labels can obtain visual explanations of deep neural network such as attention maps by back-propagating gradients.
Compositionality of semantic concepts in image synthesis and analysis is appealing as it can help in decomposing known and generatively recomposing unknown data.
Therefore, the source-domain task of interest solution (e. g. a classifier for classification tasks) which is jointly trained with the source-domain representation can be applicable to both the source and target representations.
First, we show through psychovisual studies that different people have different emotional reactions to the same image, which is a strong and novel departure from previous work that only records and predicts a single dominant emotion for each image.