Abstract: In this work, we investigate exemplar-free class incremental learning (CIL) with knowledge distillation (KD) as a regularization strategy, aiming to prevent forgetting. KD-based methods are successfully used in CIL, but they often struggle to regularize the model without access to exemplars of the training data from previous tasks. Our analysis reveals that this issue originates from substantial representation shifts in the teacher network when dealing with out-of-distribution data. This causes large errors in the KD loss component, leading to performance degradation in CIL. Inspired by recent test-time adaptation methods, we introduce Teacher Adaptation (TA), a method that concurrently updates the teacher and the main model during incremental training. Our method seamlessly integrates with KD-based CIL approaches and allows for consistent enhancement of their performance across multiple exemplar-free CIL benchmarks.

Abstract: Researchers have extensively studied the field of vision and language, discovering that both visual and textual content is crucial for understanding scenes effectively. Particularly, comprehending text in videos holds great significance, requiring both scene text understanding and temporal reasoning. This paper focuses on exploring two recently introduced datasets, NewsVideoQA and M4-ViteVQA, which aim to address video question answering based on textual content. The NewsVideoQA dataset contains question-answer pairs related to the text in news videos, while M4- ViteVQA comprises question-answer pairs from diverse categories like vlogging, traveling, and shopping. We provide an analysis of the formulation of these datasets on various levels, exploring the degree of visual understanding and multi-frame comprehension required for answering the questions. Additionally, the study includes experimentation with BERT-QA, a text-only model, which demonstrates comparable performance to the original methods on both datasets, indicating the shortcomings in the formulation of these datasets. Furthermore, we also look into the domain adaptation aspect by examining the effectiveness of training on M4-ViteVQA and evaluating on NewsVideoQA and vice-versa, thereby shedding light on the challenges and potential benefits of out-of-domain training.

Abstract: We call on the Document AI (DocAI) community to re-evaluate current methodologies and embrace the challenge of creating more practically-oriented benchmarks. Document Understanding Dataset and Evaluation (DUDE) seeks to remediate the halted research progress in understanding visually-rich documents (VRDs). We present a new dataset with novelties related to types of questions, answers, and document layouts based on multi-industry, multi-domain, and multi-page VRDs of various origins and dates. Moreover, we are pushing the boundaries of current methods by creating multi-task and multi-domain evaluation setups that more accurately simulate real-world situations where powerful generalization and adaptation under low-resource settings are desired. DUDE aims to set a new standard as a more practical, long-standing benchmark for the community, and we hope that it will lead to future extensions and contributions that address real-world challenges. Finally, our work illustrates the importance of finding more efficient ways to model language, images, and layout in DocAI.

Abstract: Uncertainty-based deep learning models have attracted a great deal of interest for their ability to provide accurate and reliable predictions. Evidential deep learning stands out achieving remarkable performance in detecting out-ofdistribution (OOD) data with a single deterministic neural network. Motivated by this fact, in this paper we propose the integration of an evidential deep learning method into a continual learning framework in order to perform simultaneously incremental object classification and OOD detection. Moreover, we analyze the ability of vacuity and dissonance to differentiate between in-distribution data belonging to old classes and OOD data. The proposed method 1, called CEDL, is evaluated on CIFAR-100 considering two settings consisting of 5 and 10 tasks, respectively. From the obtained results, we could appreciate that the proposed method, in addition to provide comparable results in object classification with respect to the baseline, largely outperforms OOD detection compared to several posthoc methods on three evaluation metrics: AUROC, AUPR and FPR95.

Abstract: The problem of texture segmentation and interpretation is one of the main concerns in the field of document analysis. Graphical documents often contain areas characterized by a structural texture whose recognition allows both the document understanding, and its storage in a more compact way. In this work, we focus on structural linial textures of regular repetition contained in plan documents. Starting from an atributed graph which represents the vectorized input image, we develop a method to segment textured areas and recognize their placement rules. We wish to emphasize that the searched textures do not follow a predefined pattern. Minimal closed loops of the input graph are computed, and then hierarchically clustered. In this hierarchical clustering, a distance function between two closed loops is defined in terms of their areas difference and boundary resemblance computed by a string matching procedure. Finally it is noted that, when the texture consists of isolated primitive elements, the same method can be used after computing a Voronoi Tesselation of the input graph.

Abstract: Best student paper award.
Pain detection using facial images is of critical importance in many Health applications. Since pain is a spatiotemporal process, recent works on this topic employ facial spatiotemporal features to detect pain. These systems extract such features from the entire area of the face. In this paper, we show that by employing super-pixels we can divide the face into three regions, in a way that only one of these regions (about one third of the face) contributes to the pain estimation and the other two regions can be discarded. The experimental results on the UNBCMcMaster database show that the proposed system using this single region outperforms state-of-the-art systems in detecting no-pain scenarios, while it reaches comparable results in detecting weak and severe pain scenarios.