Abstract: Continuous unsupervised representation learning (CURL) research has greatly benefited from improvements in self-supervised learning (SSL) techniques. As a result, existing CURL methods using SSL can learn high-quality representations without any labels, but with a notable performance drop when learning on a many-tasks data stream. We hypothesize that this is caused by the regularization losses that are imposed to prevent forgetting, leading to a suboptimal plasticity-stability trade-off: they either do not adapt fully to the incoming data (low plasticity), or incur significant forgetting when allowed to fully adapt to a new SSL pretext-task (low stability). In this work, we propose to train an expert network that is relieved of the duty of keeping the previous knowledge and can focus on performing optimally on the new tasks (optimizing plasticity). In the second phase, we combine this new knowledge with the previous network in an adaptation-retrospection phase to avoid forgetting and initialize a new expert with the knowledge of the old network. We perform several experiments showing that our proposed approach outperforms other CURL exemplar-free methods in few- and many-task split settings. Furthermore, we show how to adapt our approach to semi-supervised continual learning (Semi-SCL) and show that we surpass the accuracy of other exemplar-free Semi-SCL methods and reach the results of some others that use exemplars.

Abstract: When used in a real-world noisy environment, the capacity to generalize to multiple domains is essential for any autonomous scene text spotting system. However, existing state-of-the-art methods employ pretraining and fine-tuning strategies on natural scene datasets, which do not exploit the feature interaction across other complex domains. In this work, we explore and investigate the problem of domain-agnostic scene text spotting, i.e., training a model on multi-domain source data such that it can directly generalize to target domains rather than being specialized for a specific domain or scenario. In this regard, we present the community a text spotting validation benchmark called Under-Water Text (UWT) for noisy underwater scenes to establish an important case study. Moreover, we also design an efficient super-resolution based end-to-end transformer baseline called DA-TextSpotter which achieves comparable or superior performance over existing text spotting architectures for both regular and arbitrary-shaped scene text spotting benchmarks in terms of both accuracy and model efficiency. The dataset, code and pre-trained models will be released upon acceptance.

Abstract: Object detection in documents is a key step to automate the structural elements identification process in a digital or scanned document through understanding the hierarchical structure and relationships between different elements. Large and complex models, while achieving high accuracy, can be computationally expensive and memory-intensive, making them impractical for deployment on resource constrained devices. Knowledge distillation allows us to create small and more efficient models that retain much of the performance of their larger counterparts. Here we present a graph-based knowledge distillation framework to correctly identify and localize the document objects in a document image. Here, we design a structured graph with nodes containing proposal-level features and edges representing the relationship between the different proposal regions. Also, to reduce text bias an adaptive node sampling strategy is designed to prune the weight distribution and put more weightage on non-text nodes. We encode the complete graph as a knowledge representation and transfer it from the teacher to the student through the proposed distillation loss by effectively capturing both local and global information concurrently. Extensive experimentation on competitive benchmarks demonstrates that the proposed framework outperforms the current state-of-the-art approaches. The code will be available at: this https URL.

Abstract: Historical cryptology studies (original) encrypted manuscripts, often handwritten sources, produced in our history. These historical sources can be found in archives, often hidden without any indexing and therefore hard to locate. Once found they need to be digitized and turned into a machine-readable text format before they can be deciphered with computational methods. The focus of historical cryptology is not primarily the development of sophisticated algorithms for decipherment, but rather the entire process of analysis of the encrypted source from collection and digitization to transcription and decryption. The process also includes the interpretation and contextualization of the message set in its historical context. There are many challenges on the way, such as mistakes made by the scribe, errors made by the transcriber, damaged pages, handwriting styles that are difficult to interpret, historical languages from various time periods, and hidden underlying language of the message. Ciphertexts vary greatly in terms of their code system and symbol sets used with more or less distinguishable symbols. Ciphertexts can be embedded in clearly written text, or shorter or longer sequences of cleartext can be embedded in the ciphertext. The ciphers used mostly in historical times are substitutions (simple, homophonic, or polyphonic), with or without nomenclatures, encoded as digits or symbol sequences, with or without spaces. So the circumstances are different from those in modern cryptography which focuses on methods (algorithms) and their strengths and assumes that the algorithm is applied correctly. For both historical and modern cryptology, attack vectors outside the algorithm are applied like implementation flaws and side-channel attacks. In this chapter, we give an introduction to the field of historical cryptology and present an overview of how researchers today process historical encrypted sources.

Abstract: Conditional human motion generation is an important topic with many applications in virtual reality, gaming, and robotics. While prior works have focused on generating motion guided by text, music, or scenes, these typically result in isolated motions confined to short durations. Instead, we address the generation of long, continuous sequences guided by a series of varying textual descriptions. In this context, we introduce FlowMDM, the first diffusion-based model that generates seamless Human Motion Compositions (HMC) without any postprocessing or redundant denoising steps. For this, we introduce the Blended Positional Encodings, a technique that leverages both absolute and relative positional encodings in the denoising chain. More specifically, global motion coherence is recovered at the absolute stage, whereas smooth and realistic transitions are built at the relative stage. As a result, we achieve state-of-the-art results in terms of accuracy, realism, and smoothness on the Babel and HumanML3D datasets. FlowMDM excels when trained with only a single description per motion sequence thanks to its Pose-Centric Cross-ATtention, which makes it robust against varying text descriptions at inference time. Finally, to address the limitations of existing HMC metrics, we propose two new metrics: the Peak Jerk and the Area Under the Jerk, to detect abrupt transitions.

Abstract: Generating high-fidelity 3D-aware images without 3D supervision is a valuable capability in various applications. Current methods based on NeRF features, SDF information, or triplane features have limited variation after training. To address this, we propose a novel approach that combines pretrained models for shape and content generation. Our method leverages a pretrained Neural Radiance Field as a shape prior and a diffusion model for content generation. By conditioning the diffusion model with 3D features, we enhance its ability to generate novel views with 3D awareness. We introduce a consistency token shared between the NeRF module and the diffusion model to maintain 3D consistency during sampling. Moreover, our framework allows for text editing of 3D-aware image generation, enabling users to modify the style over 3D views while preserving semantic content. Our contributions include incorporating 3D awareness into a text-to-image model, addressing identity consistency in 3D view synthesis, and enabling text editing of 3D-aware image generation. We provide detailed explanations, including the shape prior based on the NeRF model and the content generation process using the diffusion model. We also discuss challenges such as shape consistency and sampling saturation. Experimental results demonstrate the effectiveness and visual quality of our approach.

Abstract: RGB cloth generation has been deeply studied in the related literature, however, 3D garment generation remains an open problem. In this paper, we build a conditional variational autoencoder for 3D garment generation and draping. We propose a pyramid network to add garment details progressively in a canonical space, i.e. unposing and unshaping the garments w.r.t. the body. We study conditioning the network on surface normal UV maps, as an intermediate representation, which is an easier problem to optimize than 3D coordinates. Our results on two public datasets, CLOTH3D and CAPE, show that our model is robust, controllable in terms of detail generation by the use of multi-resolution pyramids, and achieves state-of-the-art results that can highly generalize to unseen garments, poses, and shapes even when training with small amounts of data.

Abstract: rogress in lighting estimation is tracked by computing existing image quality assessment (IQA) metrics on images from standard datasets. While this may appear to be a reasonable approach, we demonstrate that doing so does not correlate to human preference when the estimated lighting is used to relight a virtual scene into a real photograph. To study this, we design a controlled psychophysical experiment where human observers must choose their preference amongst rendered scenes lit using a set of lighting estimation algorithms selected from the recent literature, and use it to analyse how these algorithms perform according to human perception. Then, we demonstrate that none of the most popular IQA metrics from the literature, taken individually, correctly represent human perception. Finally, we show that by learning a combination of existing IQA metrics, we can more accurately represent human preference. This provides a new perceptual framework to help evaluate future lighting estimation algorithms.