Abstract: The goal of the challenge is to develop a test-time adaptation (TTA) method, which could adapt the model to gradually changing domains in video sequences for semantic segmentation task. It is based on a synthetic driving video dataset – SHIFT. The source model is trained on images taken during daytime in clear weather. Domain changes at test-time are mainly caused by varying weather conditions and times of day. The TTA methods are evaluated in each image sequence (video) separately, meaning the model is reset to the source model state before the next sequence. Images come one by one and a prediction has to be made at the arrival of each frame. Each sequence is composed of 401 images and starts with the source domain, then gradually drifts to a different one (changing weather or time of day) until the middle of the sequence. In the second half of the sequence, the domain gradually shifts back to the source one. Ground truth data is available only for the validation split of the SHIFT dataset, in which there are only six sequences that start and end with the source domain. We conduct an analysis specifically on those sequences. Ground truth data for test split, on which the developed TTA methods are evaluated for leader board ranking, are not publicly available.
The proposed solution secured a 3rd place in a challenge and received an innovation award. Contrary to the solutions that scored better, we did not use any external pretrained models or specialized data augmentations, to keep the solutions as general as possible. We have focused on analyzing the distributional shift and developing a method that could adapt to changing data dynamics and generalize across different scenarios.

Abstract: The field of visual document understanding has witnessed a rapid growth in emerging challenges and powerful multi-modal strategies. However, they rely on an extensive amount of document data to learn their pretext objectives in a ``pre-train-then-fine-tune'' paradigm and thus, suffer a significant performance drop in real-world online industrial settings. One major reason is the over-reliance on OCR engines to extract local positional information within a document page. Therefore, this hinders the model's generalizability, flexibility and robustness due to the lack of capturing global information within a document image. We introduce TransferDoc, a cross-modal transformer-based architecture pre-trained in a self-supervised fashion using three novel pretext objectives. TransferDoc learns richer semantic concepts by unifying language and visual representations, which enables the production of more transferable models. Besides, two novel downstream tasks have been introduced for a ``closer-to-real'' industrial evaluation scenario where TransferDoc outperforms other state-of-the-art approaches.

Abstract: This manuscript presents a review of state-of-the-art techniques proposed in the literature for multimodal image registration, addressing instances where images from different modalities need to be precisely aligned in the same reference system. This scenario arises when the images to be registered come from different modalities, among the visible and thermal spectral bands, 3D-RGB, or flash-no flash, or NIR-visible. The review spans different techniques from classical approaches to more modern ones based on deep learning, aiming to highlight the particularities required at each step in the registration pipeline when dealing with multimodal images. It is noteworthy that medical images are excluded from this review due to their specific characteristics, including the use of both active and passive sensors or the non-rigid nature of the body contained in the image.

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.

Abstract: Remote sensing applications, impacted by acquisition season and sensor variety, require high-resolution images. Transformer-based models improve satellite image super-resolution but are less effective than convolutional neural networks (CNNs) at extracting local details, crucial for image clarity. This paper introduces SWViT-RRDB, a new deep learning model for satellite imagery super-resolution. The SWViT-RRDB, combining transformer with convolution and attention blocks, overcomes the limitations of existing models by better representing small objects in satellite images. In this model, a pipeline of residual fusion group (RFG) blocks is used to combine the multi-headed self-attention (MSA) with residual in residual dense block (RRDB). This combines global and local image data for better super-resolution. Additionally, an overlapping cross-attention block (OCAB) is used to enhance fusion and allow interaction between neighboring pixels to maintain long-range pixel dependencies across the image. The SWViT-RRDB model and its larger variants outperform state-of-the-art (SoTA) models on two different satellite datasets in terms of PSNR and SSIM.

Abstract: Deep learning has made significant advances in recent years, and as a result, it is now in a stage where it can achieve outstanding results in tasks requiring visual understanding of scenes. However, its performance tends to decline when dealing with low-quality images. The advent of super-resolution (SR) techniques has started to have an impact on the field of remote sensing by enabling the restoration of fine details and enhancing image quality, which could help to increase performance in other vision tasks. However, in previous works, contradictory results for scene visual understanding were achieved when SR techniques were applied. In this paper, we present an experimental study on the impact of SR on enhancing aerial scene classification. Through the analysis of different state-of-the-art SR algorithms, including traditional methods and deep learning-based approaches, we unveil the transformative potential of SR in overcoming the limitations of low-resolution (LR) aerial imagery. By enhancing spatial resolution, more fine details are captured, opening the door for an improvement in scene understanding. We also discuss the effect of different image scales on the quality of SR and its effect on aerial scene classification. Our experimental work demonstrates the significant impact of SR on enhancing aerial scene classification compared to LR images, opening new avenues for improved remote sensing applications.

Abstract: This work proposes a novel approach that integrates super-resolution techniques with off-the-shelf object detection methods to tackle the problem of handling very low-resolution thermal images. The suggested approach begins by enhancing the low-resolution (LR) thermal images through a guided super-resolution strategy, leveraging a high-resolution (HR) visible spectrum image. Subsequently, object detection is performed on the high-resolution thermal image. The experimental results demonstrate tremendous improvements in comparison with both scenarios: when object detection is performed on the LR thermal image alone, as well as when object detection is conducted on the up-sampled LR thermal image. Moreover, the proposed approach proves highly valuable in camouflaged scenarios where objects might remain undetected in visible spectrum images.

Abstract: Document Visual Question Answering (DocVQA) is a fast growing branch of document understanding. Despite the fact that documents contain sensitive or copyrighted information, none of the current DocVQA methods offers strong privacy guarantees.
In this work, we explore privacy in the domain of DocVQA for the first time. We highlight privacy issues in state of the art multi-modal LLM models used for DocVQA, and explore possible solutions.
Specifically, we focus on the invoice processing use case as a realistic, widely used scenario for document understanding, and propose a large scale DocVQA dataset comprising invoice documents and associated questions and answers. We employ a federated learning scheme, that reflects the real-life distribution of documents in different businesses, and we explore the use case where the ID of the invoice issuer is the sensitive information to be protected.
We demonstrate that non-private models tend to memorise, behaviour that can lead to exposing private information. We then evaluate baseline training schemes employing federated learning and differential privacy in this multi-modal scenario, where the sensitive information might be exposed through any of the two input modalities: vision (document image) or language (OCR tokens).
Finally, we design an attack exploiting the memorisation effect of the model, and demonstrate its effectiveness in probing different DocVQA models.