Publications by Lorenzo Baraldi

Action Detection is a complex task that aims to detect and classify human actions in video clips. Typically, it has been addressed by processing fine-grained features extracted from a video classification backbone. Recently, thanks to the robustness of object and people detectors, a deeper focus has been added on relationship modelling. Following this line, we propose a graph-based framework to learn high-level interactions between people and objects, in both space and time. In our formulation, spatio-temporal relationships are learned through self-attention on a multi-layer graph structure which can connect entities from consecutive clips, thus considering long-range spatial and temporal dependencies. The proposed module is backbone independent by design and does not require end-to-end training. Extensive experiments are conducted on the AVA dataset, where our model demonstrates state-of-the-art results and consistent improvements over baselines built with different backbones. Code is publicly available at https://github.com/aimagelab/STAGE_action_detection.

Handwritten Text Recognition (HTR) in free-layout pages is a valuable yet challenging task which aims to automatically understand handwritten texts. State-of-the-art approaches in this field usually encode input images with Convolutional Neural Networks, whose kernels are typically defined on a fixed grid and focus on all input pixels independently. However, this is in contrast with the sparse nature of handwritten pages, in which only pixels representing the ink of the writing are useful for the recognition task. Furthermore, the standard convolution operator is not explicitly designed to take into account the great variability in shape, scale, and orientation of handwritten characters. To overcome these limitations, we investigate the use of deformable convolutions for handwriting recognition. This type of convolution deform the convolution kernel according to the content of the neighborhood, and can therefore be more adaptable to geometric variations and other deformations of the text. Experiments conducted on the IAM and RIMES datasets demonstrate that the use of deformable convolutions is a promising direction for the design of novel architectures for handwritten text recognition.

Recurrent Neural Networks with Long Short-Term Memory (LSTM) make use of gating mechanisms to mitigate exploding and vanishing gradients when learning long-term dependencies. For this reason, LSTMs and other gated RNNs are widely adopted, being the standard de facto for many sequence modeling tasks. Although the memory cell inside the LSTM contains essential information, it is not allowed to influence the gating mechanism directly. In this work, we improve the gate potential by including information coming from the internal cell state. The proposed modification, named Working Memory Connection, consists in adding a learnable nonlinear projection of the cell content into the network gates. This modification can fit into the classical LSTM gates without any assumption on the underlying task, being particularly effective when dealing with longer sequences. Previous research effort in this direction, which goes back to the early 2000s, could not bring a consistent improvement over vanilla LSTM. As part of this paper, we identify a key issue tied to previous connections that heavily limits their effectiveness, hence preventing a successful integration of the knowledge coming from the internal cell state. We show through extensive experimental evaluation that Working Memory Connections constantly improve the performance of LSTMs on a variety of tasks. Numerical results suggest that the cell state contains useful information that is worth including in the gate structure.

Text-image retrieval has been recently becoming a hot-spot research field, thanks to the development of deeply-learnable architectures which can retrieve visual items given textual queries and vice-versa. The key idea of many state-of-the-art approaches has been that of learning a joint multi-modal embedding space in which text and images could be projected and compared. Here we take a different approach and reformulate the problem of text-image retrieval as that of learning a translation between the textual and visual domain. Our proposal leverages an end-to-end trainable architecture that can translate text into image features and vice versa and regularizes this mapping with a cycle-consistency criterion. Experimental evaluations for text-to-image and image-to-text retrieval, conducted on small, medium and large-scale datasets show consistent improvements over the baselines, thus confirming the appropriateness of using a cycle-consistent constrain for the text-image matching task.

Augmented reality (AR) is the process of using technology to superimpose images, text or sounds on top of what a person can already see. Art galleries and museums started to develop AR applications to increase engagement and provide an entirely new kind of exploration experience. However, the creation of contents results a very time consuming process, thus requiring an ad-hoc development for each painting to be increased. In fact, for the creation of an AR experience on any painting, it is necessary to choose the points of interest, to create digital content and then to develop the application. If this is affordable for the great masterpieces of an art gallery, it would be impracticable for an entire collection. In this context, the idea of this paper is to develop AR applications based on Artificial Intelligence. In particular, automatic captioning techniques are the key core for the implementation of AR application for improving the user experience in front of a painting or an artwork in general. The study has demonstrated the feasibility through a proof of concept application, implemented for hand held devices, and adds to the body of knowledge in mobile AR application as this approach has not been applied in this field before.

Replicating the human ability to connect Vision and Language has recently been gaining a lot of attention in the Computer Vision and the Natural Language Processing communities. This research effort has resulted in algorithms that can retrieve images from textual descriptions and vice versa, when realistic images and sentences with simple semantics are employed and when paired training data is provided. In this paper, we go beyond these limitations and tackle the design of visual-semantic algorithms in the domain of the Digital Humanities. This setting not only advertises more complex visual and semantic structures but also features a significant lack of training data which makes the use of fully-supervised approaches infeasible. With this aim, we propose a joint visual-semantic embedding that can automatically align illustrations and textual elements without paired supervision. This is achieved by transferring the knowledge learned on ordinary visual-semantic datasets to the artistic domain. Experiments, performed on two datasets specifically designed for this domain, validate the proposed strategies and quantify the domain shift between natural images and artworks.

Transformer-based architectures represent the state of the art in sequence modeling tasks like machine translation and language understanding. Their applicability to multi-modal contexts like image captioning, however, is still largely under-explored. With the aim of filling this gap, we present M² - a Meshed Transformer with Memory for Image Captioning. The architecture improves both the image encoding and the language generation steps: it learns a multi-level representation of the relationships between image regions integrating learned a priori knowledge, and uses a mesh-like connectivity at decoding stage to exploit low- and high-level features. Experimentally, we investigate the performance of the M² Transformer and different fully-attentive models in comparison with recurrent ones. When tested on COCO, our proposal achieves a new state of the art in single-model and ensemble configurations on the "Karpathy" test split and on the online test server. We also assess its performances when describing objects unseen in the training set. Trained models and code for reproducing the experiments are publicly available at :https://github.com/aimagelab/meshed-memory-transformer.

The ability to generate natural language explanations conditioned on the visual perception is a crucial step towards autonomous agents which can explain themselves and communicate with humans. While the research efforts in image and video captioning are giving promising results, this is often done at the expense of the computational requirements of the approaches, limiting their applicability to real contexts. In this paper, we propose a fully-attentive captioning algorithm which can provide state-of-the-art performances on language generation while restricting its computational demands. Our model is inspired by the Transformer model and employs only two Transformer layers in the encoding and decoding stages. Further, it incorporates a novel memory-aware encoding of image regions. Experiments demonstrate that our approach achieves competitive results in terms of caption quality while featuring reduced computational demands. Further, to evaluate its applicability on autonomous agents, we conduct experiments on simulated scenes taken from the perspective of domestic robots.

Connected Components Labeling is an essential step of many Image Processing and Computer Vision tasks. Since the first proposal of a labeling algorithm, which dates back to the sixties, many approaches have optimized the computational load needed to label an image. In particular, the use of decision forests and state prediction have recently appeared as valuable strategies to improve performance. However, due to the overhead of the manual construction of prediction states and the size of the resulting machine code, the application of these strategies has been restricted to small masks, thus ignoring the benefit of using a block-based approach. In this paper, we combine a block-based mask with state prediction and code compression: the resulting algorithm is modeled as a Directed Rooted Acyclic Graph with multiple entry points, which is automatically generated without manual intervention. When tested on synthetic and real datasets, in comparison with optimized implementations of state-of-the-art algorithms, the proposed approach shows superior performance, surpassing the results obtained by all compared approaches in all settings.