Publications

Open-vocabulary semantic segmentation aims at segmenting arbitrary categories expressed in textual form. Previous works have trained over large amounts of image-caption pairs to enforce pixel-level multimodal alignments. However captions provide global information about the semantics of a given image but lack direct localization of individual concepts. Further training on large-scale datasets inevitably brings significant computational costs. In this paper we propose FreeDA a training-free diffusion-augmented method for open-vocabulary semantic segmentation which leverages the ability of diffusion models to visually localize generated concepts and local-global similarities to match class-agnostic regions with semantic classes. Our approach involves an offline stage in which textual-visual reference embeddings are collected starting from a large set of captions and leveraging visual and semantic contexts. At test time these are queried to support the visual matching process which is carried out by jointly considering class-agnostic regions and global semantic similarities. Extensive analyses demonstrate that FreeDA achieves state-of-the-art performance on five datasets surpassing previous methods by more than 7.0 average points in terms of mIoU and without requiring any training. Our source code is available at https://aimagelab.github.io/freeda/.

The failures of Deep Networks can sometimes be ascribed to biases in the data or algorithmic choices. Existing debiasing approaches exploit prior knowledge to avoid unintended solutions; we acknowledge that, in real-world settings, it could be unfeasible to gather enough prior information to characterize the bias, or it could even raise ethical considerations. We hence propose a novel debiasing approach, termed ClusterFix, which does not require any external hint about the nature of biases. Such an approach alters the standard empirical risk minimization and introduces a per-example weight, encoding how critical and far from the majority an example is. Notably, the weights consider how difficult it is for the model to infer the correct pseudo-label, which is obtained in a self-supervised manner by dividing examples into multiple clusters. Extensive experiments show that the misclassification error incurred in identifying the correct cluster allows for identifying examples prone to bias-related issues. As a result, our approach outperforms existing methods on standard benchmarks for bias removal and fairness.

The objective of image captioning models is to bridge the gap between the visual and linguistic modalities by generating natural language descriptions that accurately reflect the content of input images. In recent years, researchers have leveraged deep learning-based models and made advances in the extraction of visual features and the design of multimodal connections to tackle this task. This work presents a novel approach toward developing image captioning models that utilize an external kNN memory to improve the generation process. Specifically, we propose two model variants that incorporate a knowledge retriever component that is based on visual similarities, a differentiable encoder to represent input images, and a kNN-augmented language model to predict tokens based on contextual cues and text retrieved from the external memory. We experimentally validate our approach on COCO and nocaps datasets and demonstrate that incorporating an explicit external memory can significantly enhance the quality of captions, especially with a larger retrieval corpus. This work provides valuable insights into retrieval-augmented captioning models and opens up new avenues for improving image captioning at a larger scale.

In response to the rising threat of the face morphing attack, this paper introduces and explores the potential of Video-based Morphing Attack Detection (V-MAD) systems in real-world operational scenarios. While current morphing attack detection methods primarily focus on a single or a pair of images, V-MAD is based on video sequences, exploiting the video streams acquired by face verification tools available, for instance, at airport gates. We show for the first time the advantages that the availability of multiple probe frames brings to the morphing attack detection task, especially in scenarios where the quality of probe images is varied. Experimental results on a real operational database demonstrate that video sequences represent valuable information for increasing the performance of morphing attack detection systems.

End-to-end transformer-based trackers have achieved remarkable performance on most human-related datasets. However, training these trackers in heterogeneous scenarios poses significant challenges, including negative interference - where the model learns conflicting scene-specific parameters - and limited domain generalization, which often necessitates expensive fine-tuning to adapt the models to new domains. In response to these challenges, we introduce Parameter-efficient Scenario-specific Tracking Architecture (PASTA), a novel framework that combines Parameter-Efficient Fine-Tuning (PEFT) and Modular Deep Learning (MDL). Specifically, we define key scenario attributes (e.g, camera-viewpoint, lighting condition) and train specialized PEFT modules for each attribute. These expert modules are combined in parameter space, enabling systematic generalization to new domains without increasing inference time. Extensive experiments on MOTSynth, along with zero-shot evaluations on MOT17 and PersonPath22 demonstrate that a neural tracker built from carefully selected modules surpasses its monolithic counterpart. We release models and code.

The conventional training approach for image captioning involves pre-training a network using teacher forcing and subsequent fine-tuning with Self-Critical Sequence Training to maximize hand-crafted captioning metrics. However, when attempting to optimize modern and higher-quality metrics like CLIP-Score and PAC-Score, this training method often encounters instability and fails to acquire the genuine descriptive capabilities needed to produce fluent and informative captions. In this paper, we propose a new training paradigm termed Direct CLIP-Based Optimization (DiCO). Our approach jointly learns and optimizes a reward model that is distilled from a learnable captioning evaluator with high human correlation. This is done by solving a weighted classification problem directly inside the captioner. At the same time, DiCO prevents divergence from the original model, ensuring that fluency is maintained. DiCO not only exhibits improved stability and enhanced quality in the generated captions but also aligns more closely with human preferences compared to existing methods, especially in modern metrics. Additionally, it maintains competitive performance in traditional metrics.

Through the Face Morphing attack is possible to use the same legal document by two different people, destroying the unique biometric link between the document and its owner. In other words, a morphed face image has the potential to bypass face verification-based security controls, then representing a severe security threat. Unfortunately, the lack of public, extensive and varied training datasets severely hampers the development of effective and robust Morphing Attack Detection (MAD) models, key tools in contrasting the Face Morphing attack since able to automatically detect the presence of morphing images. Indeed, privacy regulations limit the possibility of acquiring, storing, and transferring MAD-related data that contain personal information, such as faces. Therefore, in this paper, we investigate the use of Federated Learning to train a MAD model on local training samples across multiple sites, eliminating the need for a single centralized training dataset, as common in Machine Learning, and then overcoming privacy limitations. Experimental results suggest that FL is a viable solution that will need to be considered in future research works in MAD.

Implicit biases, subtle and unconscious attitudes, permeate various facets of human decision-making and are similarly pervasive in Artificial Intelligence (AI) systems. These biases can stem from shortcut learning, where models rely on superficial patterns that do not capture the underlying phenomena. Inspired by social psychology studies, we introduce two novel metrics to analyze implicit biases in visual-language models. Our comprehensive analysis of 90 open-clip models reveals widespread anomalies related to ethnicity and gender. The first metric considers the cosine similarity between images and text prompts related to social stereotypes. The second metric adapts the Implicit Association Test (IAT), which evaluates prejudice and hidden discrimination within human behavior. Our findings illustrate that conventional text-based debiasing efforts can inadvertently amplify second-order biases instead of mitigating them. Furthermore, in expanding our evaluation to multimodal Large Language Models (LLMs), we demonstrate disparities in the tendency to generate semantically positive or negative outputs, depending on the ethnicity or gender of the individuals depicted in the input images.