In federated learning, textual prompt tuning adapts Vision-Language Models (e.g., CLIP) by tuning lightweight input tokens (or prompts) on local client data, while keeping network weights frozen. After training, only the prompts are shared by the clients with the central server for aggregation. However, textual prompt tuning suffers from overfitting to known concepts, limiting its generalizability to unseen concepts. To address this limitation, we propose Multimodal Visual Prompt Tuning (FedMVP) that conditions the prompts on multimodal contextual information - derived from the input image and textual attribute features of a class. At the core of FedMVP is a PromptFormer module that synergistically aligns textual and visual features through a cross-attention mechanism. The dynamically generated multimodal visual prompts are then input to the frozen vision encoder of CLIP, and trained with a combination of CLIP similarity loss and a consistency loss. Extensive evaluation on 20 datasets, spanning three generalization settings, demonstrates that FedMVP not only preserves performance on in-distribution classes and domains, but also displays higher generalizability to unseen classes and domains, surpassing state-of-the-art methods by a notable margin of +1.57% - 2.26%. Code is available at https://github.com/mainaksingha01/FedMVP.

Real-world datasets typically exhibit long-tailed (LT) distributions, where a few head classes dominate and many tail classes are severely underrepresented. While recent work shows that parameter-efficient fine-tuning (PEFT) methods like LoRA and AdaptFormer preserve tail-class performance on foundation models such as CLIP, we find that they do so at the cost of head-class accuracy. We identify the head-tail ratio, the proportion of head to tail classes, as a crucial but overlooked factor influencing this trade-off. Through controlled experiments on CIFAR100 with varying imbalance ratio ($\rho$) and head-tail ratio ($\eta$), we show that PEFT excels in tail-heavy scenarios but degrades in more balanced and head-heavy distributions. To overcome these limitations, we propose LT-Soups, a two-stage model soups framework designed to generalize across diverse LT regimes. In the first stage, LT-Soups averages models fine-tuned on balanced subsets to reduce head-class bias; in the second, it fine-tunes only the classifier on the full dataset to restore head-class accuracy. Experiments across six benchmark datasets show that LT-Soups achieves superior trade-offs compared to both PEFT and traditional model soups across a wide range of imbalance regimes.

Domain Generalized Semantic Segmentation (DGSS) deals with training a model on a labeled source domain with the aim of generalizing to unseen domains during inference. Existing DGSS methods typically effectuate robust features by means of Domain Randomization (DR). Such an approach is often limited as it can only account for style diversification and not content. In this work, we take an orthogonal approach to DGSS and propose to use an assembly of CoLlaborative FOUndation models for Domain Generalized Semantic Segmentation (CLOUDS). In detail, CLOUDS is a framework that integrates Foundation Models of various kinds: (i) CLIP backbone for its robust feature representation, (ii) Diffusion Model to diversify the content, thereby covering various modes of the possible target distribution, and (iii) Segment Anything Model (SAM) for iteratively refining the predictions of the segmentation model. Extensive experiments show that our CLOUDS excels in adapting from synthetic to real DGSS benchmarks and under varying weather conditions, notably outperforming prior methods by 5.6% and 6.7% on averaged mIoU, respectively. Our code is available at https://github.com/yasserben/CLOUDS

Identifying subordinate-level categories from images is a longstanding task in computer vision and is referred to as fine-grained visual recognition (FGVR). It has tremendous significance in real-world applications since an average layperson does not excel at differentiating species of birds or mushrooms due to subtle differences among the species. A major bottleneck in developing FGVR systems is caused by the need of high-quality paired expert annotations. To circumvent the need of expert knowledge we propose Fine-grained Semantic Category Reasoning (FineR) that internally leverages the world knowledge of large language models (LLMs) as a proxy in order to reason about fine-grained category names. In detail, to bridge the modality gap between images and LLM, we extract part-level visual attributes from images as text and feed that information to a LLM. Based on the visual attributes and its internal world knowledge the LLM reasons about the subordinate-level category names. Our training-free FineR outperforms several state-of-the-art FGVR and language and vision assistant models and shows promise in working in the wild and in new domains where gathering expert annotation is arduous.

Dengue fever is mostly prevalent in tropical and subtropical regions, where Aedes mosquitoes, the primary vectors for the virus, thrive in warm and humid environments. In West Bengal, a province in India, the typical duration of the dengue disease spans two to three months, necessitating substantial infrastructure for dengue patients during this period. If the government heavily invests in developing this infrastructure, there's a risk of these facilities remaining underutilized during periods of low dengue incidence. Conversely, without adequate investment, dengue could potentially escalate into an epidemic. This research seeks to identify regions where insufficient infrastructure impedes public healthcare systems from catering to dengue patients. The primary focus of this paper is to evaluate the necessary degree of public-private collaborations needed to address seasonal dengue epidemics and pinpoint specific durations within the healthcare system that require attention.

Open-set Unsupervised Video Domain Adaptation (OU-VDA) deals with the task of adapting an action recognition model from a labelled source domain to an unlabelled target domain that contains “target-private” categories, which are present in the target but absent in the source. In this work we deviate from the prior work of training a specialized open-set classifier or weighted adversarial learning by proposing to use pre-trained Language and Vision Models (CLIP). The CLIP is well suited for OUVDA due to its rich representation and the zero-shot recognition capabilities. However, rejecting target-private instances with the CLIP's zero-shot protocol requires oracle knowledge about the target-private label names. To circumvent the impossibility of the knowledge of label names, we propose AutoLabel that automatically discovers and generates object-centric compositional candidate target-private class names. Despite its simplicity, we show that CLIP when equipped with AutoLabel can satisfactorily reject the target-private instances, thereby facilitating better alignment between the shared classes of the two domains. The code is available11https://github.com/gzaraunitn/autolabel.

Source-Free Video Unsupervised Domain Adaptation (SFVUDA) task consists in adapting an action recognition model, trained on a labelled source dataset, to an unlabelled target dataset, without accessing the actual source data. The previous approaches have attempted to address SFVUDA by leveraging self-supervision (e.g., enforcing temporal consistency) derived from the target data itself. In this work, we take an orthogonal approach by exploiting "web-supervision" from Large Language-Vision Models (LLVMs), driven by the rationale that LLVMs contain a rich world prior surprisingly robust to domain-shift. We showcase the unreasonable effectiveness of integrating LLVMs for SFVUDA by devising an intuitive and parameter-efficient method, which we name Domain Adaptation with Large Language-Vision models (DALL-V), that distills the world prior and complementary source model information into a student network tailored for the target. Despite the simplicity, DALL-V1achieves significant improvement over state-of-the-art SFVUDA methods.

In this work we address multi-target domain adaptation (MTDA) in semantic segmentation, which consists in adapting a single model from an annotated source dataset to multiple unannotated target datasets that differ in their underlying data distributions. To address MTDA, we propose a self-training strategy that employs pseudo-labels to induce cooperation among multiple domain-specific classifiers. We employ feature stylization as an efficient way to generate image views that forms an integral part of self-training. Additionally, to prevent the network from overfitting to noisy pseudo-labels, we devise a rectification strategy that leverages the predictions from different classifiers to estimate the quality of pseudo-labels. Our extensive experiments on numerous settings, based on four different semantic segmentation datasets, validates the effectiveness of the proposed self-training strategy and shows that our method outperforms state-of-the-art MTDA approaches. https://github.com/Mael-zys/CoaST.

In this paper we address multi-target domain adaptation (MTDA), where given one labeled source dataset and multiple unlabeled target datasets that differ in data distributions, the task is to learn a robust predictor for all the target domains. We identify two key aspects that can help to alleviate multiple domain-shifts in the MTDA: feature aggregation and curriculum learning. To this end, we propose Curriculum Graph Co-Teaching (CGCT) that uses a dual classifier head, with one of them being a graph convolutional network (GCN) which aggregates features from similar samples across the domains. To prevent the classifiers from over-fitting on its own noisy pseudo-labels we develop a co-teaching strategy with the dual classifier head that is assisted by curriculum learning to obtain more reliable pseudo-labels. Furthermore, when the domain labels are available, we propose Domain-aware Curriculum Learning (DCL), a sequential adaptation strategy that first adapts on the easier target domains, followed by the harder ones. We experimentally demonstrate the effectiveness of our proposed frameworks on several benchmarks and advance the state-of-the-art in the MTDA by large margins (e.g. +5.6% on the DomainNet).

In this paper, we address Novel Class Discovery (NCD), the task of unveiling new classes in a set of unlabeled samples given a labeled dataset with known classes. We exploit the peculiarities of NCD to build a new framework, named Neighborhood Contrastive Learning (NCL), to learn discriminative representations that are important to clustering performance. Our contribution is twofold. First, we find that a feature extractor trained on the labeled set generates representations in which a generic query sample and its neighbors are likely to share the same class. We exploit this observation to retrieve and aggregate pseudo-positive pairs with contrastive learning, thus encouraging the model to learn more discriminative representations. Second, we notice that most of the instances are easily discriminated by the network, contributing less to the contrastive loss. To overcome this issue, we propose to generate hard negatives by mixing labeled and unlabeled samples in the feature space. We experimentally demonstrate that these two ingredients significantly contribute to clustering performance and lead our model to outperform state-of-the-art methods by a large margin (e.g., clustering accuracy +13% on CIFAR-100 and +8% on ImageNet).

Hashing methods have recently been shown to be very effective in the retrieval of remote sensing (RS) images due to their computational efficiency and fast search speed. Common hashing methods in RS are based on hand-crafted features on top of which they learn a hash function, which provides the final binary codes. However, these features are not optimized for the final task (i.e., retrieval using binary codes). On the other hand, modern deep neural networks (DNNs) have shown an impressive success in learning optimized features for a specific task in an end-to-end fashion. Unfortunately, typical RS data sets are composed of only a small number of labeled samples, which make the training (or fine-tuning) of big DNNs problematic and prone to overfitting. To address this problem, in this letter, we introduce a metric-learning-based hashing network, which: 1) implicitly uses a big, pretrained DNN as an intermediate representation step without the need of retraining or fine-tuning; 2) learns a semantic-based metric space where the features are optimized for the target retrieval task; and 3) computes compact binary hash codes for fast search. Experiments carried out on two RS benchmarks highlight that the proposed network significantly improves the retrieval performance under the same retrieval time when compared to the state-of-the-art hashing methods in RS.

Abstract Most domain adaptation methods consider the problem of transferring knowledge to the target domain from a single-source dataset. However, in practical applications, we typically have access to multiple sources. In this paper we propose the first approach for multi-source domain adaptation (MSDA) based on generative adversarial networks. Our method is inspired by the observation that the appearance of a given image depends on three factors: thedomain, thestyle(characterized in terms of low-level features variations) and thecontent. For this reason, we propose to project the source image features onto a space where only the dependence from the content is kept, and then re-project this invariant representation onto the pixel space using the target domain and style. In this way, new labeled images can be generated which are used to train a final target classifier. We test our approach using common MSDA benchmarks, showing that it outperforms state-of-the-art methods.

Recent co-part segmentation methods mostly operate in a supervised learning setting, which requires a large amount of annotated data for training. To overcome this limitation, we propose a self-supervised deep learning method for co-part segmentation. Differently from previous works, our approach develops the idea that motion information inferred from videos can be leveraged to discover meaningful object parts. To this end, our method relies on pairs of frames sampled from the same video. The network learns to predict part segments together with a representation of the motion between two frames, which permits reconstruction of the target image. Through extensive experimental evaluation on publicly available video sequences we demonstrate that our approach can produce improved segmentation maps with respect to previous self-supervised co-part segmentation approaches.

Deep learning (DL) has proved successful in medical imaging and, in the wake of the recent COVID-19 pandemic, some works have started to investigate DL-based solutions for the assisted diagnosis of lung diseases. While existing works focus on CT scans, this paper studies the application of DL techniques for the analysis of lung ultrasonography (LUS) images. Specifically, we present a novel fully-annotated dataset of LUS images collected from several Italian hospitals, with labels indicating the degree of disease severity at a frame-level, video-level, and pixel-level (segmentation masks). Leveraging these data, we introduce several deep models that address relevant tasks for the automatic analysis of LUS images. In particular, we present a novel deep network, derived from Spatial Transformer Networks, which simultaneously predicts the disease severity score associated to a input frame and provides localization of pathological artefacts in a weakly-supervised way. Furthermore, we introduce a new method based on uninorms for effective frame score aggregation at a video-level. Finally, we benchmark state of the art deep models for estimating pixel-level segmentations of COVID-19 imaging biomarkers. Experiments on the proposed dataset demonstrate satisfactory results on all the considered tasks, paving the way to future research on DL for the assisted diagnosis of COVID-19 from LUS data.

A classifier trained on a dataset seldom works on other datasets obtained under different conditions due to domain shift. This problem is commonly addressed by domain adaptation methods. In this work we introduce a novel deep learning framework which unifies different paradigms in unsupervised domain adaptation. Specifically, we propose domain alignment layers which implement feature whitening for the purpose of matching source and target feature distributions. Additionally, we leverage the unlabeled target data by proposing the Min-Entropy Consensus loss, which regularizes training while avoiding the adoption of many user-defined hyper-parameters. We report results on publicly available datasets, considering both digit classification and object recognition tasks. We show that, in most of our experiments, our approach improves upon previous methods, setting new state-of-the-art performances.

Most of the public satellite image datasets contain only a small number of annotated images. The lack of a sufficient quantity of labeled data for training is a bottleneck for the use of modern deep-learning based classification approaches in this domain. In this paper we propose a semi -supervised approach to deal with this problem. We use the discriminator (D) of a Generative Adversarial Network (GAN) as the final classifier, and we train D using both labeled and unlabeled data. The main novelty we introduce is the representation of the visual information fed to D by means of two different channels: the original image and its “semantic” representation, the latter being obtained by means of an external network trained on ImageNet. The two channels are fused in D and jointly used to classify fake images, real labeled and real unlabeled images. We show that using only 100 labeled images, the proposed approach achieves an accuracy close to 69% and a significant improvement with respect to other GAN-based semi-supervised methods. Although we have tested our approach only on satellite images, we do not use any domain-specific knowledge. Thus, our method can be applied to other semi-supervised domains.

The growing volume of Remote Sensing (RS) image archives demands for feature learning techniques and hashing functions which can: (1) accurately represent the semantics in the RS images; and (2) have quasi real-time performance during retrieval. This paper aims to address both challenges at the same time, by learning a semantic-based metric space for content based RS image retrieval while simultaneously producing binary hash codes for an efficient archive search. This double goal is achieved by training a deep network using a combination of different loss functions which, on the one hand, aim at clustering semantically similar samples (i.e., images), and, on the other hand, encourage the network to produce final activation values (i.e., descriptors) that can be easily binarized. Moreover, since RS annotated training images are too few to train a deep network from scratch, we propose to split the image representation problem in two different phases. In the first we use a general-purpose, pre-trained network to produce an intermediate representation, and in the second we train our hashing network using a relatively small set of training images. Experiments on two aerial benchmark archives show that the proposed method outperforms previous state-of-the-art hashing approaches by up to 5.4% using the same number of hash bits per image.