Abstract: This book compiles leading research on the development of explainable and interpretable machine learning methods in the context of computer vision and machine learning.
Research progress in computer vision and pattern recognition has led to a variety of modeling techniques with almost human-like performance. Although these models have obtained astounding results, they are limited in their explainability and interpretability: what is the rationale behind the decision made? what in the model structure explains its functioning? Hence, while good performance is a critical required characteristic for learning machines, explainability and interpretability capabilities are needed to take learning machines to the next step to include them in decision support systems involving human supervision.
This book, written by leading international researchers, addresses key topics of explainability and interpretability, including the following:

·Evaluation and Generalization in Interpretable Machine Learning
·Explanation Methods in Deep Learning
·Learning Functional Causal Models with Generative Neural Networks
·Learning Interpreatable Rules for Multi-Label Classification
·Structuring Neural Networks for More Explainable Predictions
·Generating Post Hoc Rationales of Deep Visual Classification Decisions
·Ensembling Visual Explanations
·Explainable Deep Driving by Visualizing Causal Attention
·Interdisciplinary Perspective on Algorithmic Job Candidate Search
·Multimodal Personality Trait Analysis for Explainable Modeling of Job Interview Decisions
·Inherent Explainability Pattern Theory-based Video Event Interpretations

Abstract: Sketch-RGB cross-modality person re-identification (ReID) is a challenging task that aims to match a sketch portrait drawn by a professional artist with a full-body photo taken by surveillance equipment to deal with situations where the monitoring equipment is damaged at the accident scene. However, sketch portraits only provide highly abstract frontal body contour information and lack other important features such as color, pose, behavior, etc. The difference in saliency between the two modalities brings new challenges to cross-modality person ReID. To overcome this problem, this paper proposes a novel dual-stream model for cross-modality person ReID, which is able to mine modality-invariant features to reduce the discrepancy between sketch and camera images end-to-end. More specifically, we propose a multi-spectrum image fusion (MSIF) method, which aims to exploit the image appearance changes brought by multiple spectrums and guide the network to mine modality-invariant commonalities during training. It only processes the spectrum of the input images without adding additional calculations and model complexity, which can be easily integrated into other models. Moreover, we introduce a joint structure via a generalized mean pooling (GMP) layer and a self-attention (SA) mechanism to balance background and texture information and obtain the regional features with a large amount of information in the image. To further shrink the intra-class distance, a weighted regularized triplet (WRT) loss is developed without introducing additional hyperparameters. The model was first evaluated on the PKU Sketch ReID dataset, and extensive experimental results show that the Rank-1/mAP accuracy of our method is 87.00%/91.12%, reaching the current state-of-the-art performance. To further validate the effectiveness of our approach in handling cross-modality person ReID, we conducted experiments on two commonly used IR-RGB datasets (SYSU-MM01 and RegDB). The obtained results show that our method achieves competitive performance. These results confirm the ability of our method to effectively process images from different modalities.

Abstract: This paper addresses the problem of inferring unseen cross-modal image-to-image translations between multiple modalities. We assume that only some of the pairwise translations have been seen (i.e. trained) and infer the remaining unseen translations (where training pairs are not available). We propose mix and match networks, an approach where multiple encoders and decoders are aligned in such a way that the desired translation can be obtained by simply cascading the source encoder and the target decoder, even when they have not interacted during the training stage (i.e. unseen). The main challenge lies in the alignment of the latent representations at the bottlenecks of encoder-decoder pairs. We propose an architecture with several tools to encourage alignment, including autoencoders and robust side information and latent consistency losses. We show the benefits of our approach in terms of effectiveness and scalability compared with other pairwise image-to-image translation approaches. We also propose zero-pair cross-modal image translation, a challenging setting where the objective is inferring semantic segmentation from depth (and vice-versa) without explicit segmentation-depth pairs, and only from two (disjoint) segmentation-RGB and depth-RGB training sets. We observe that a certain part of the shared information between unseen modalities might not be reachable, so we further propose a variant that leverages pseudo-pairs which allows us to exploit this shared information between the unseen modalities

Abstract: This work presents and evaluates a novel compact scene representation based on Stixels that infers geometric and semantic information. Our approach overcomes the previous rather restrictive geometric assumptions for Stixels by introducing a novel depth model to account for non-flat roads and slanted objects. Both semantic and depth cues are used jointly to infer the scene representation in a sound global energy minimization formulation. Furthermore, a novel approximation scheme is introduced in order to significantly reduce the computational complexity of the Stixel algorithm, and then achieve real-time computation capabilities. The idea is to first perform an over-segmentation of the image, discarding the unlikely Stixel cuts, and apply the algorithm only on the remaining Stixel cuts. This work presents a novel over-segmentation strategy based on a fully convolutional network, which outperforms an approach based on using local extrema of the disparity map. We evaluate the proposed methods in terms of semantic and geometric accuracy as well as run-time on four publicly available benchmark datasets. Our approach maintains accuracy on flat road scene datasets while improving substantially on a novel non-flat road dataset.

Abstract: Edges are key components of any visual scene to the extent that we can recognise objects merely by their silhouettes. The human visual system captures edge information through neurons in the visual cortex that are sensitive to both intensity discontinuities and particular orientations. The “classical approach” assumes that these cells are only responsive to the stimulus present within their receptive fields, however, recent studies demonstrate that surrounding regions and inter-areal feedback connections influence their responses significantly. In this work we propose a biologically-inspired edge detection model in which orientation selective neurons are represented through the first derivative of a Gaussian function resembling double-opponent cells in the primary visual cortex (V1). In our model we account for four kinds of receptive field surround, i.e. full, far, iso- and orthogonal-orientation, whose contributions are contrast-dependant. The output signal from V1 is pooled in its perpendicular direction by larger V2 neurons employing a contrast-variant centre-surround kernel. We further introduce a feedback connection from higher-level visual areas to the lower ones. The results of our model on three benchmark datasets show a big improvement compared to the current non-learning and biologically-inspired state-of-the-art algorithms while being competitive to the learning-based methods.

Abstract: Gesture Recognition (GR) is a challenging research area in computer vision. To tackle the annotation bottleneck in GR, we formulate the problem of Zero-Shot Gesture Recognition (ZS-GR) and propose a two-stream model from two input modalities: RGB and Depth videos. To benefit from the vision Transformer capabilities, we use two vision Transformer models, for human detection and visual features representation. We configure a transformer encoder-decoder architecture, as a fast and accurate human detection model, to overcome the challenges of the current human detection models. Considering the human keypoints, the detected human body is segmented into nine parts. A spatio-temporal representation from human body is obtained using a vision Transformer and a LSTM network. A semantic space maps the visual features to the lingual embedding of the class labels via a Bidirectional Encoder Representations from Transformers (BERT) model. We evaluated the proposed model on five datasets, Montalbano II, MSR Daily Activity 3D, CAD-60, NTU-60, and isoGD obtaining state-of-the-art results compared to state-of-the-art ZS-GR models as well as the Zero-Shot Action Recognition (ZS-AR).