Publications

Convolutional Neural Networks (CNNs) have been broadly employed in dermoscopic image analysis, mainly as a result of the large amount of data gathered by the International Skin Imaging Collaboration (ISIC). Like in many other medical imaging domains, state-of-the-art methods take advantage of architectures developed for other tasks, frequently assuming full transferability between enormous sets of natural images (eg{} ImageNet) and dermoscopic images, which is not always the case. With this paper we provide a comprehensive analysis on the effectiveness of state-of-the-art deep learning techniques when applied to dermoscopic image analysis. In order to achieve this goal, we consider several CNNs architectures and analyze how their performance is affected by the size of the network, image resolution, data augmentation process, amount of available data, and model calibration. Moreover, taking advantage of the analysis performed, we design a novel ensemble method to further increase the classification accuracy. The proposed solution achieved the third best result in the 2019 official ISIC challenge, with an accuracy of 0.593.

Face morphing and related morphing attacks have emerged as a serious security threat for automatic face recognition systems and a challenging research field. Therefore, the availability of effective and reliable morphing attack detectors is strongly needed. In this paper, we proposed a framework based on a double Siamese architecture to tackle the morphing attack detection task in the differential scenario, in which two images, a trusted live acquired image and a probe image (morphed or bona fide) are given as the input for the system. In particular, the presented framework aimed to merge the information computed by two different modules to predict the final score. The first one was designed to extract information about the identity of the input faces, while the second module was focused on the detection of artifacts related to the morphing process. Experimental results were obtained through several and rigorous cross-dataset tests, exploiting three well-known datasets, namely PMDB, MorphDB, and AMSL, containing automatic and manually refined facial morphed images, showing that the proposed framework was able to achieve satisfying results.

Connected Components Labeling represents a fundamental step for many Computer Vision and Image Processing pipelines. Since the first appearance of the task in the sixties, many algorithmic solutions to optimize the computational load needed to label an image have been proposed. Among them, block-based scan approaches and decision trees revealed to be some of the most valuable strategies. However, due to the cost of the manual construction of optimal decision trees and the computational limitations of automatic strategies employed in the past, the application of blocks and decision trees has been restricted to small masks, and thus to 2D algorithms. With this paper we present a novel heuristic algorithm based on decision tree learning methodology, called Entropy Partitioning Decision Tree (EPDT). It allows to compute near-optimal decision trees for large scan masks. Experimental results demonstrate that algorithms based on the generated decision trees outperform state-of-the-art competitors.

This paper provides a detailed description of how to install, setup, and use the YACCLAB benchmark to test the algorithms published in "A Heuristic-Based Decision Tree for Connected Components Labeling of 3D Volumes," underlying how the parameters affect and influence experimental results.

The joint understanding of vision and language has been recently gaining a lot of attention in both the Computer Vision and Natural Language Processing communities, with the emergence of tasks such as image captioning, image-text matching, and visual question answering. As both images and text can be encoded as sets or sequences of elements - like regions and words - proper reduction functions are needed to transform a set of encoded elements into a single response, like a classification or similarity score. In this paper, we propose a novel fully-attentive reduction method for vision and language. Specifically, our approach computes a set of scores for each element of each modality employing a novel variant of cross-attention, and performs a learnable and cross-modal reduction, which can be used for both classification and ranking. We test our approach on image-text matching and visual question answering, building fair comparisons with other reduction choices, on both COCO and VQA 2.0 datasets. Experimentally, we demonstrate that our approach leads to a performance increase on both tasks. Further, we conduct ablation studies to validate the role of each component of the approach.

Traditionally, the analysis of histological samples is visually performed by a pathologist, who inspects under the microscope the tissue samples, looking for malignancies and anomalies. This visual assessment is both time consuming and highly unreliable due to the subjectivity of the evaluation. Hence, there are growing efforts towards the automatisation of such analysis, oriented to the development of computer-aided diagnostic tools, with a ever-growing role of techniques based on deep learning. In this work, we analyze some of the issues commonly associated with providing deep learning based techniques to medical professionals. We thus introduce a tool, aimed at both researchers and medical professionals, which simplifies and accelerates the training and exploitation of such models. The outcome of the tool is an attention map representing cancer probability distribution on top of the Whole Slide Image, driving the pathologist through a faster and more accurate diagnostic procedure.

Anticipating human motion in crowded scenarios is essential for developing intelligent transportation systems, social-aware robots and advanced video surveillance applications. A key component of this task is represented by the inherently multi-modal nature of human paths which makes socially acceptable multiple futures when human interactions are involved. To this end, we propose a generative architecture for multi-future trajectory predictions based on Conditional Variational Recurrent Neural Networks (C-VRNNs). Conditioning mainly relies on prior belief maps, representing most likely moving directions and forcing the model to consider past observed dynamics in generating future positions. Human interactions are modelled with a graph-based attention mechanism enabling an online attentive hidden state refinement of the recurrent estimation. To corroborate our model, we perform extensive experiments on publicly-available datasets (e.g., ETH/UCY, Stanford Drone Dataset, STATS SportVU NBA, Intersection Drone Dataset and TrajNet++) and demonstrate its effectiveness in crowded scenes compared to several state-of-the-art methods.

In this paper, we address the problem of generating person images conditioned on both pose and appearance information. Specifically, given an image xa of a person and a target pose P(xb), extracted from an image xb, we synthesize a new image of that person in pose P(xb), while preserving the visual details in xa. In order to deal with pixel-to-pixel misalignments caused by the pose differences between P(xa) and P(xb), we introduce deformable skip connections in the generator of our Generative Adversarial Network. Moreover, a nearest-neighbour loss is proposed instead of the common L1 and L2 losses in order to match the details of the generated image with the target image. Quantitative and qualitative results, using common datasets and protocols recently proposed for this task, show that our approach is competitive with respect to the state of the art. Moreover, we conduct an extensive evaluation using off-the-shell person re-identification (Re-ID) systems trained with person-generation based augmented data, which is one of themain important applications for this task. Our experiments show that our Deformable GANs can significantly boost the Re-ID accuracy and are even better than data-augmentation methods specifically trained using Re-ID losses.