Abstract: Most works on image retrieval from text queries have addressed the problem of retrieving semantically relevant images. However, the ability to assess the aesthetic quality of an image is an increasingly important differentiating factor for search engines. In this work, given a semantic query, we are interested in retrieving images which are semantically relevant and score highly in terms of aesthetics/visual quality. We use large-margin classifiers and rankers to learn statistical models capable of ordering images based on the aesthetic and semantic information. In particular, we compare two families of approaches: while the first one attempts to learn a single ranker which takes into account both semantic and aesthetic information, the second one learns separate semantic and aesthetic models. We carry out a quantitative and qualitative evaluation on a recently-published large-scale dataset and we show that the second family of techniques significantly outperforms the first one.

Abstract: Clothing is still manually manufactured for the most part nowadays, resulting in discrepancies between nominal and real dimensions, and potentially ill-fitting garments. Hence, it is common in the apparel industry to manually perform measures at preshipment time. We present an automatic method to obtain such measures from a single image of a garment that speeds up this task. It is generic and extensible in the sense that it does not depend explicitly on the garment shape or type. Instead, it learns through a probabilistic graphical model to identify the different contour parts. Subsequently, a set of Lasso regressors, one per desired measure, can predict the actual values of the measures. We present results on a dataset of 130 images of jackets and 98 of pants, of varying sizes and styles, obtaining 1.17 and 1.22 cm of mean absolute error, respectively.

Abstract: In this paper we propose to learn a multimodal image and text embedding from Web and Social Media data, aiming to leverage the semantic knowledge learnt in the text domain and transfer it to a visual model for semantic image retrieval. We demonstrate that the pipeline can learn from images with associated text without supervision and perform a thourough analysis of five different text embeddings in three different benchmarks. We show that the embeddings learnt with Web and Social Media data have competitive performances over supervised methods in the text based image retrieval task, and we clearly outperform state of the art in the MIRFlickr dataset when training in the target data. Further we demonstrate how semantic multimodal image retrieval can be performed using the learnt embeddings, going beyond classical instance-level retrieval problems. Finally, we present a new dataset, InstaCities1M, composed by Instagram images and their associated texts that can be used for fair comparison of image-text embeddings.

Abstract: Learning to count is a learning strategy that has been recently proposed in the literature for dealing with problems where estimating the number of object instances in a scene is the final objective. In this framework, the task of learning to detect and localize individual object instances is seen as a harder task that can be evaded by casting the problem as that of computing a regression value from hand-crafted image features. In this paper we explore the features that are learned when training a counting convolutional neural
network in order to understand their underlying representation.
To this end we define a counting problem for MNIST data and show that the internal representation of the network is able to classify digits in spite of the fact that no direct supervision was provided for them during training.
We also present preliminary results about a deep network that is able to count the number of pedestrians in a scene.

Abstract: arXiv:2001.08311
Recent advances in unsupervised domain adaptation have shown the effectiveness of adversarial training to adapt features across domains, endowing neural networks with the capability of being tested on a target domain without requiring any training annotations in this domain. The great majority of existing domain adaptation models rely on image translation networks, which often contain a huge amount of domain-specific parameters. Additionally, the feature adaptation step often happens globally, at a coarse level, hindering its applicability to tasks such as semantic segmentation, where details are of crucial importance to provide sharp results. In this thesis, we present a novel architecture, which learns to adapt features across domains by taking into account per class information. To that aim, we design a conditional pixel-wise discriminator network, whose output is conditioned on the segmentation masks. Moreover, following recent advances in image translation, we adopt the recently introduced StarGAN architecture as image translation backbone, since it is able to perform translations across multiple domains by means of a single generator network. Preliminary results on a segmentation task designed to assess the effectiveness of the proposed approach highlight the potential of the model, improving upon strong baselines and alternative designs.

Abstract: IntraVascular UltraSound (IVUS) is a technique allowing the diagnosis of coronary plaque. An accurate (semi-)automatic assessment of the luminal contours could speed up the diagnosis. In most of the approaches, the information on the vessel shape is obtained combining a supervised learning step with a local refinement algorithm. In this paper, we explore for the first time, the use of a Convolutional Neural Networks (CNN) architecture that on one hand is able to extract the optimal image features and at the same time can serve as a supervised classifier to detect the lumen border in IVUS images. The main limitation of CNN, relies on the fact that this technique requires a large amount of training data due to the huge amount of parameters that it has. To
solve this issue, we introduce a patch classification approach to generate an extended training-set from a few annotated images. An accuracy of 93% and F-score of 71% was obtained with this technique, even when it was applied to challenging frames containig calcified plaques, stents and catheter shadows.

Abstract: We propose an amelioration of proposed Graph Embedding (GEM) method in previous work that takes advantages of structural pattern representation and the structured distortion. it models an Attributed Graph (AG) as a Probabilistic Graphical Model (PGM). Then, it learns the parameters of this PGM presented by a vector, as new signature of AG in a lower dimensional vectorial space. We focus to adapt the structured learning algorithm via 1_slack formulation with a suitable risk function, called Graph Edit Distance (GED). It defines the dissimilarity of the ground truth and predicted graph labels. It determines by the error tolerant graph matching using bipartite graph matching algorithm. We apply Structured Support Vector Machines (SSVM) to process classification task. During our experiments, we got our results on the GREC dataset.

Abstract: Major advances have recently been made in merging language and vision representations. Most tasks considered so far have confined themselves to the processing of objects and lexicalised relations amongst objects (content words). We know, however, that humans (even pre-school children) can abstract over raw multimodal data to perform certain types of higher level reasoning, expressed in natural language by function words. A case in point is given by their ability to learn quantifiers, i.e. expressions like few, some and all. From formal semantics and cognitive linguistics, we know that quantifiers are relations over sets which, as a simplification, we can see as proportions. For instance, in most fish are red, most encodes the proportion of fish which are red fish. In this paper, we study how well current neural network strategies model such relations. We propose a task where, given an image and a query expressed by an object–property pair, the system must return a quantifier expressing which proportions of the queried object have the queried property. Our contributions are twofold. First, we show that the best performance on this task involves coupling state-of-the-art attention mechanisms with a network architecture mirroring the logical structure assigned to quantifiers by classic linguistic formalisation. Second, we introduce a new balanced dataset of image scenarios associated with quantification queries, which we hope will foster further research in this area.