Laurent Najman

Quentin Garrido, Laurent Najman, Yann Lecun

Recent progress has been made towards learning invariant or equivariant representations with self-supervised learning. While invariant methods are evaluated on large scale datasets, equivariant ones are evaluated in smaller, more controlled, settings. We aim at bridging the gap between the two in order to learn more diverse representations that are suitable for a wide range of tasks. We start by introducing a dataset called 3DIEBench, consisting of renderings from 3D models over 55 classes and more than 2.5 million images where we have full control on the transformations applied to the objects. We further introduce a predictor architecture based on hypernetworks to learn equivariant representations with no possible collapse to invariance. We introduce SIE (Split Invariant-Equivariant) which combines the hypernetwork-based predictor with representations split in two parts, one invariant, the other equivariant, to learn richer representations. We demonstrate significant performance gains over existing methods on equivariance related tasks from both a qualitative and quantitative point of view. We further analyze our introduced predictor and show how it steers the learned latent space. We hope that both our introduced dataset and approach will enable learning richer representations without supervision in more complex scenarios. Code and data are available at https://github.com/facebookresearch/SIE.

Quentin Garrido, Yubei Chen, Adrien Bardes et al.

Recent approaches in self-supervised learning of image representations can be categorized into different families of methods and, in particular, can be divided into contrastive and non-contrastive approaches. While differences between the two families have been thoroughly discussed to motivate new approaches, we focus more on the theoretical similarities between them. By designing contrastive and covariance based non-contrastive criteria that can be related algebraically and shown to be equivalent under limited assumptions, we show how close those families can be. We further study popular methods and introduce variations of them, allowing us to relate this theoretical result to current practices and show the influence (or lack thereof) of design choices on downstream performance. Motivated by our equivalence result, we investigate the low performance of SimCLR and show how it can match VICReg's with careful hyperparameter tuning, improving significantly over known baselines. We also challenge the popular assumption that non-contrastive methods need large output dimensions. Our theoretical and quantitative results suggest that the numerical gaps between contrastive and non-contrastive methods in certain regimes can be closed given better network design choices and hyperparameter tuning. The evidence shows that unifying different SOTA methods is an important direction to build a better understanding of self-supervised learning.

Quentin Garrido, Randall Balestriero, Laurent Najman et al.

Joint-Embedding Self Supervised Learning (JE-SSL) has seen a rapid development, with the emergence of many method variations but only few principled guidelines that would help practitioners to successfully deploy them. The main reason for that pitfall comes from JE-SSL's core principle of not employing any input reconstruction therefore lacking visual cues of unsuccessful training. Adding non informative loss values to that, it becomes difficult to deploy SSL on a new dataset for which no labels can help to judge the quality of the learned representation. In this study, we develop a simple unsupervised criterion that is indicative of the quality of the learned JE-SSL representations: their effective rank. Albeit simple and computationally friendly, this method -- coined RankMe -- allows one to assess the performance of JE-SSL representations, even on different downstream datasets, without requiring any labels. A further benefit of RankMe is that it does not have any training or hyper-parameters to tune. Through thorough empirical experiments involving hundreds of training episodes, we demonstrate how RankMe can be used for hyperparameter selection with nearly no reduction in final performance compared to the current selection method that involve a dataset's labels. We hope that RankMe will facilitate the deployment of JE-SSL towards domains that do not have the opportunity to rely on labels for representations' quality assessment.

Zeev Gutman, Ritvik Vij, Laurent Najman et al.

Current approaches to generic segmentation start by creating a hierarchy of nested image partitions and then specifying a segmentation from it. Our first contribution is to describe several ways, most of them new, for specifying segmentations using the hierarchy elements. Then, we consider the best hierarchy-induced segmentation specified by a limited number of hierarchy elements. We focus on a common quality measure for binary segmentations, the Jaccard index (also known as IoU). Optimizing the Jaccard index is highly non-trivial, and yet we propose an efficient approach for doing exactly that. This way we get algorithm-independent upper bounds on the quality of any segmentation created from the hierarchy. We found that the obtainable segmentation quality varies significantly depending on the way that the segments are specified by the hierarchy elements, and that representing a segmentation with only a few hierarchy elements is often possible. (Code is available).

Nicolas Boutry, Laurent Najman, Thierry Géraud

In Mathematical Morphology (MM), connected filters based on dynamics are used to filter the extrema of an image. Similarly, persistence is a concept coming from Persistent Homology (PH) and Morse Theory (MT) that represents the stability of the extrema of a Morse function. Since these two concepts seem to be closely related, in this paper we examine their relationship, and we prove that they are equal on n-D Morse functions, n $\ge$ 1. More exactly, pairing a minimum with a 1-saddle by dynamics or pairing the same 1-saddle with a minimum by persistence leads exactly to the same pairing, assuming that the critical values of the studied Morse function are unique. This result is a step further to show how much topological data analysis and mathematical morphology are related, paving the way for a more in-depth study of the relations between these two research fields.

Caroline Mazini Rodrigues, Nicolas Boutry, Laurent Najman

Providing interpretability of deep-learning models to non-experts, while fundamental for a responsible real-world usage, is challenging. Attribution maps from xAI techniques, such as Integrated Gradients, are a typical example of a visualization technique containing a high level of information, but with difficult interpretation. In this paper, we propose two methods, Maximum Activation Groups Extraction (MAGE) and Multiscale Interpretable Visualization (Ms-IV), to explain the model's decision, enhancing global interpretability. MAGE finds, for a given CNN, combinations of features which, globally, form a semantic meaning, that we call concepts. We group these similar feature patterns by clustering in ``concepts'', that we visualize through Ms-IV. This last method is inspired by Occlusion and Sensitivity analysis (incorporating causality), and uses a novel metric, called Class-aware Order Correlation (CaOC), to globally evaluate the most important image regions according to the model's decision space. We compare our approach to xAI methods such as LIME and Integrated Gradients. Experimental results evince the Ms-IV higher localization and faithfulness values. Finally, qualitative evaluation of combined MAGE and Ms-IV demonstrates humans' ability to agree, based on the visualization, with the decision of clusters' concepts; and, to detect, among a given set of networks, the existence of bias.

Mariia Zameshina, Olivier Teytaud, Fabien Teytaud et al.

We design general-purpose algorithms for addressing fairness issues and mode collapse in generative modeling. More precisely, to design fair algorithms for as many sensitive variables as possible, including variables we might not be aware of, we assume no prior knowledge of sensitive variables: our algorithms use unsupervised fairness only, meaning no information related to the sensitive variables is used for our fairness-improving methods. All images of faces (even generated ones) have been removed to mitigate legal risks.

Sarah Almeida Carneiro, Giovanni Chierchia, Jean Charléty et al.

Although traffic is one of the massively collected data, it is often only available for specific regions. One concern is that, although there are studies that give good results for these data, the data from these regions may not be sufficiently representative to describe all the traffic patterns in the rest of the world. In quest of addressing this concern, we propose a speed prediction method that is independent of large historical speed data. To predict a vehicle's speed, we use the trajectory road topographical features to fit a Shared Weight Multilayer Perceptron learning model. Our results show significant improvement, both qualitative and quantitative, over standard regression analysis. Moreover, the proposed framework sheds new light on the way to design new approaches for traffic analysis.

Ismail Labiad, Thomas Bäck, Pierre Fernandez et al.

In many cases, adversarial attacks are based on specialized algorithms specifically dedicated to attacking automatic image classifiers. These algorithms perform well, thanks to an excellent ad hoc distribution of initial attacks. However, these attacks are easily detected due to their specific initial distribution. We therefore consider other black-box attacks, inspired from generic black-box optimization tools, and in particular the log-normal algorithm. We apply the log-normal method to the attack of fake detectors, and get successful attacks: importantly, these attacks are not detected by detectors specialized on classical adversarial attacks. Then, combining these attacks and deep detection, we create improved fake detectors.

Léo Hein, Giovanni de Nunzio, Giovanni Chierchia et al.

Existing traffic volume estimation methods typically address either forecasting traffic on sensor-equipped roads or spatially imputing missing volumes using nearby sensors. While forecasting models generally disregard unmonitored roads by design, spatial imputation methods explicitly address network-wide estimation; yet this approach relies on volume data at inference time, limiting its applicability in sensor-scarce cities. Unlike traffic volume data, probe vehicle speeds and static road attributes are more broadly accessible and support full coverage of road segments in most urban networks. In this work, we present the Hybrid Directed-Attention Spatio-Temporal Graph Neural Network (HDA-STGNN), an inductive deep learning framework designed to tackle the network-wide volume estimation problem. Our approach leverages speed profiles, static road attributes, and road network topology to predict daily traffic volume profiles across all road segments in the network. To evaluate the effectiveness of our approach, we perform extensive ablation studies that demonstrate the model's capacity to capture complex spatio-temporal dependencies and highlight the value of topological information for accurate network-wide traffic volume estimation without relying on volume data at inference time.

Mariia Zameshina, Olivier Teytaud, Laurent Najman

Latent diffusion models excel at producing high-quality images from text. Yet, concerns appear about the lack of diversity in the generated imagery. To tackle this, we introduce Diverse Diffusion, a method for boosting image diversity beyond gender and ethnicity, spanning into richer realms, including color diversity.Diverse Diffusion is a general unsupervised technique that can be applied to existing text-to-image models. Our approach focuses on finding vectors in the Stable Diffusion latent space that are distant from each other. We generate multiple vectors in the latent space until we find a set of vectors that meets the desired distance requirements and the required batch size.To evaluate the effectiveness of our diversity methods, we conduct experiments examining various characteristics, including color diversity, LPIPS metric, and ethnicity/gender representation in images featuring humans.The results of our experiments emphasize the significance of diversity in generating realistic and varied images, offering valuable insights for improving text-to-image models. Through the enhancement of image diversity, our approach contributes to the creation of more inclusive and representative AI-generated art.

Yuliang Gu, Yepeng Liu, Zhichao Sun et al.

Annotating 3D medical images demands expert knowledge and is time-consuming. As a result, semi-supervised learning (SSL) approaches have gained significant interest in 3D medical image segmentation. The significant size differences among various organs in the human body lead to imbalanced class distribution, which is a major challenge in the real-world application of these SSL approaches. To address this issue, we develop a novel Shape Transformation driven by Active Contour (STAC), that enlarges smaller organs to alleviate imbalanced class distribution across different organs. Inspired by curve evolution theory in active contour methods, STAC employs a signed distance function (SDF) as the level set function, to implicitly represent the shape of organs, and deforms voxels in the direction of the steepest descent of SDF (i.e., the normal vector). To ensure that the voxels far from expansion organs remain unchanged, we design an SDF-based weight function to control the degree of deformation for each voxel. We then use STAC as a data-augmentation process during the training stage. Experimental results on two benchmark datasets demonstrate that the proposed method significantly outperforms some state-of-the-art methods. Source code is publicly available at https://github.com/GuGuLL123/STAC.

Aditya Challa, Sravan Danda, B. S. Daya Sagar et al.

Hyperspectral images (HSI) consist of rich spatial and spectral information, which can potentially be used for several applications. However, noise, band correlations and high dimensionality restrict the applicability of such data. This is recently addressed using creative deep learning network architectures such as ResNet, SSRN, and A2S2K. However, the last layer, i.e the classification layer, remains unchanged and is taken to be the softmax classifier. In this article, we propose to use a watershed classifier. Watershed classifier extends the watershed operator from Mathematical Morphology for classification. In its vanilla form, the watershed classifier does not have any trainable parameters. In this article, we propose a novel approach to train deep learning networks to obtain representations suitable for the watershed classifier. The watershed classifier exploits the connectivity patterns, a characteristic of HSI datasets, for better inference. We show that exploiting such characteristics allows the Triplet-Watershed to achieve state-of-art results in supervised and semi-supervised contexts. These results are validated on Indianpines (IP), University of Pavia (UP), Kennedy Space Center (KSC) and University of Houston (UH) datasets, relying on simple convnet architecture using a quarter of parameters compared to previous state-of-the-art networks. The source code for reproducing the experiments and supplementary material (high resolution images) is available at https://github.com/ac20/TripletWatershed Code.

Quentin Garrido, Sebastian Damrich, Alexander Jäger et al.

Motivation: Single cell RNA sequencing (scRNA-seq) data makes studying the development of cells possible at unparalleled resolution. Given that many cellular differentiation processes are hierarchical, their scRNA-seq data is expected to be approximately tree-shaped in gene expression space. Inference and representation of this tree-structure in two dimensions is highly desirable for biological interpretation and exploratory analysis.Results:Our two contributions are an approach for identifying a meaningful tree structure from high-dimensional scRNA-seq data, and a visualization method respecting the tree-structure. We extract the tree structure by means of a density based minimum spanning tree on a vector quantization of the data and show that it captures biological information well. We then introduce DTAE, a tree-biased autoencoder that emphasizes the tree structure of the data in low dimensional space. We compare to other dimension reduction methods and demonstrate the success of our method both qualitatively and quantitatively on real and toy data.Availability: Our implementation relying on PyTorch and Higra is available at https://github.com/hci-unihd/DTAE.

Jord{ã}o Bragantini, Alexandre X Falc{ã}o, Laurent Najman

Despite the progress of interactive image segmentation methods, high-quality pixel-level annotation is still time-consuming and laborious - a bottleneck for several deep learning applications. We take a step back to propose interactive and simultaneous segment annotation from multiple images guided by feature space projection. This strategy is in stark contrast to existing interactive segmentation methodologies, which perform annotation in the image domain. We show that feature space annotation achieves competitive results with state-of-the-art methods in foreground segmentation datasets: iCoSeg, DAVIS, and Rooftop. Moreover, in the semantic segmentation context, it achieves 91.5% accuracy in the Cityscapes dataset, being 74.75 times faster than the original annotation procedure. Further, our contribution sheds light on a novel direction for interactive image annotation that can be integrated with existing methodologies. The supplementary material presents video demonstrations. Code available at https://github.com/LIDS-UNICAMP/rethinking-interactive-image-segmentation.

Gilles Bertrand, Laurent Najman

In the context of discrete Morse theory, we introduce Morse frames, which are maps that associate a set of critical simplexes to all simplexes. The main example of Morse frames are the Morse references. In particular, these Morse references allow computing Morse complexes, an important tool for homology. We highlight the link between Morse references and gradient flows. We also propose a novel presentation of the Annotation algorithm for persistent cohomology, as a variant of a Morse frame. Finally, we propose another construction, that takes advantage of the Morse reference for computing the Betti numbers in mod 2 arithmetic.

Mariia Zameshina, Marlene Careil, Olivier Teytaud et al.

Classical techniques for protecting facial image privacy typically fall into two categories: data-poisoning methods, exemplified by Fawkes, which introduce subtle perturbations to images, or anonymization methods that generate images resembling the original only in several characteristics, such as gender, ethnicity, or facial expression.In this study, we introduce a novel approach, PrivacyGAN, that uses the power of image generation techniques, such as VQGAN and StyleGAN, to safeguard privacy while maintaining image usability, particularly for social media applications. Drawing inspiration from Fawkes, our method entails shifting the original image within the embedding space towards a decoy image.We evaluate our approach using privacy metrics on traditional and novel facial image datasets. Additionally, we propose new criteria for evaluating the robustness of privacy-protection methods against unknown image recognition techniques, and we demonstrate that our approach is effective even in unknown embedding transfer scenarios. We also provide a human evaluation that further proves that the modified image preserves its utility as it remains recognisable as an image of the same person by friends and family.

Quentin Garrido, Mahmoud Assran, Nicolas Ballas et al.

Joint-Embedding Predictive Architecture (JEPA) has emerged as a promising self-supervised approach that learns by leveraging a world model. While previously limited to predicting missing parts of an input, we explore how to generalize the JEPA prediction task to a broader set of corruptions. We introduce Image World Models, an approach that goes beyond masked image modeling and learns to predict the effect of global photometric transformations in latent space. We study the recipe of learning performant IWMs and show that it relies on three key aspects: conditioning, prediction difficulty, and capacity. Additionally, we show that the predictive world model learned by IWM can be adapted through finetuning to solve diverse tasks; a fine-tuned IWM world model matches or surpasses the performance of previous self-supervised methods. Finally, we show that learning with an IWM allows one to control the abstraction level of the learned representations, learning invariant representations such as contrastive methods, or equivariant representations such as masked image modelling.

Quentin Garrido, Nicolas Ballas, Mahmoud Assran et al.

We investigate the emergence of intuitive physics understanding in general-purpose deep neural network models trained to predict masked regions in natural videos. Leveraging the violation-of-expectation framework, we find that video prediction models trained to predict outcomes in a learned representation space demonstrate an understanding of various intuitive physics properties, such as object permanence and shape consistency. In contrast, video prediction in pixel space and multimodal large language models, which reason through text, achieve performance closer to chance. Our comparisons of these architectures reveal that jointly learning an abstract representation space while predicting missing parts of sensory input, akin to predictive coding, is sufficient to acquire an understanding of intuitive physics, and that even models trained on one week of unique video achieve above chance performance. This challenges the idea that core knowledge -- a set of innate systems to help understand the world -- needs to be hardwired to develop an understanding of intuitive physics.

Aditya Challa, Sravan Danda, Laurent Najman

In this paper, we propose a class of non-parametric classifiers, that learn arbitrary boundaries and generalize well. Our approach is based on a novel way to regularize 1NN classifiers using a greedy approach. We refer to this class of classifiers as Watershed Classifiers. 1NN classifiers are known to trivially over-fit but have very large VC dimension, hence do not generalize well. We show that watershed classifiers can find arbitrary boundaries on any dense enough dataset, and, at the same time, have very small VC dimension; hence a watershed classifier leads to good generalization. Traditional approaches to regularize 1NN classifiers are to consider $K$ nearest neighbours. Neighbourhood component analysis (NCA) proposes a way to learn representations consistent with ($n-1$) nearest neighbour classifier, where $n$ denotes the size of the dataset. In this article, we propose a loss function which can learn representations consistent with watershed classifiers, and show that it outperforms the NCA baseline.

Sarah Almeida Carneiro, Giovanni Chierchia, Aurelie Pirayre et al.

A persistent challenge in the field of Intelligent Transportation Systems is to extract accurate traffic insights from geographic regions with scarce or no data coverage. To this end, we propose solutions for speed prediction using sparse GPS data points and their associated topographical and road design features. Our goal is to investigate whether we can use similarities in the terrain and infrastructure to train a machine learning model that can predict speed in regions where we lack transportation data. For this we create a Temporally Orientated Speed Dictionary Centered on Topographically Clustered Roads, which helps us to provide speed correlations to selected feature configurations. Our results show qualitative and quantitative improvement over new and standard regression methods. The presented framework provides a fresh perspective on devising strategies for missing data traffic analysis.

Mathurin Videau, Mariia Zameshina, Alessandro Leite et al.

AfterLearnER (After Learning Evolutionary Retrofitting) consists in applying evolutionary optimization to refine fully trained machine learning models by optimizing a set of carefully chosen parameters or hyperparameters of the model, with respect to some actual, exact, and hence possibly non-differentiable error signal, performed on a subset of the standard validation set. The efficiency of AfterLearnER is demonstrated by tackling non-differentiable signals such as threshold-based criteria in depth sensing, the word error rate in speech re-synthesis, the number of kills per life at Doom, computational accuracy or BLEU in code translation, image quality in 3D generative adversarial networks (GANs), and user feedback in image generation via Latent Diffusion Models (LDM). This retrofitting can be done after training, or dynamically at inference time by taking into account the user feedback. The advantages of AfterLearnER are its versatility, the possibility to use non-differentiable feedback, including human evaluations (i.e., no gradient is needed), the limited overfitting supported by a theoretical study, and its anytime behavior. Last but not least, AfterLearnER requires only a small amount of feedback, i.e., a few dozen to a few hundred scalars, compared to the tens of thousands needed in most related published works.

Miriam Doh, Caroline Mazini Rodrigues, Nicolas Boutry et al.

With Artificial Intelligence (AI) influencing the decision-making process of sensitive applications such as Face Verification, it is fundamental to ensure the transparency, fairness, and accountability of decisions. Although Explainable Artificial Intelligence (XAI) techniques exist to clarify AI decisions, it is equally important to provide interpretability of these decisions to humans. In this paper, we present an approach to combine computer and human vision to increase the explanation's interpretability of a face verification algorithm. In particular, we are inspired by the human perceptual process to understand how machines perceive face's human-semantic areas during face comparison tasks. We use Mediapipe, which provides a segmentation technique that identifies distinct human-semantic facial regions, enabling the machine's perception analysis. Additionally, we adapted two model-agnostic algorithms to provide human-interpretable insights into the decision-making processes.

Zelong Liu, Yuliang Gu, Zhichao Sun et al.

Crack detection is an important task in computer vision. Despite impressive in-dataset performance, deep learning-based methods still struggle in generalizing to unseen domains. The thin structure property of cracks is usually overlooked by previous methods. In this work, we introduce CrackCue, a novel method for robust crack detection based on coarse-to-fine crack cue generation. The core concept lies on leveraging the thin structure property to generate a robust crack cue, guiding the crack detection. Specifically, we first employ a simple max-pooling and upsampling operation on the crack image. This results in a coarse crack-free background, based on which a fine crack-free background can be obtained via a reconstruction network. The difference between the original image and fine crack-free background provides a fine crack cue. This fine cue embeds robust crack prior information which is unaffected by complex backgrounds, shadow, and varied lighting. As a plug-and-play method, we incorporate the proposed CrackCue into three advanced crack detection networks. Extensive experimental results demonstrate that the proposed CrackCue significantly improves the generalization ability and robustness of the baseline methods. The source code will be publicly available.

Gilson Junior Soares, Matheus Abrantes Cerqueira, Jancarlo F. Gomes et al.

Salient Object Detection (SOD) methods can locate objects that stand out in an image, assign higher values to their pixels in a saliency map, and binarize the map outputting a predicted segmentation mask. A recent tendency is to investigate pre-trained lightweight models rather than deep neural networks in SOD tasks, coping with applications under limited computational resources. In this context, we have investigated lightweight networks using a methodology named Feature Learning from Image Markers (FLIM), which assumes that the encoder's kernels can be estimated from marker pixels on discriminative regions of a few representative images. This work proposes flyweight networks, hundreds of times lighter than lightweight models, for SOD by combining a FLIM encoder with an adaptive decoder, whose weights are estimated for each input image by a given heuristic function. Such FLIM networks are trained from three to four representative images only and without backpropagation, making the models suitable for applications under labeled data constraints as well. We study five adaptive decoders; two of them are introduced here. Differently from the previous ones that rely on one neuron per pixel with shared weights, the heuristic functions of the new adaptive decoders estimate the weights of each neuron per pixel. We compare FLIM models with adaptive decoders for two challenging SOD tasks with three lightweight networks from the state-of-the-art, two FLIM networks with decoders trained by backpropagation, and one FLIM network whose labeled markers define the decoder's weights. The experiments demonstrate the advantages of the proposed networks over the baselines, revealing the importance of further investigating such methods in new applications.

Emilio Porcu, Roy El Moukari, Laurent Najman et al.

This manuscript provides a holistic (data-centric) view of what we term essential data science, as a natural ecosystem with challenges and missions stemming from the data universe with its multiple combinations of the 5D complexities (data structure, domain, cardinality, causality, and ethics) with the phases of the data life cycle. Data agents perform tasks driven by specific goals. The data scientist is an abstract entity that comes from the logical organization of data agents with their actions. Data scientists face challenges that are defined according to the missions. We define specific discipline-induced data science, which in turn allows for the definition of pan-data science, a natural ecosystem that integrates specific disciplines with the essential data science. We semantically split the essential data science into computational, and foundational. We claim that there is a serious threat of divergence between computational and foundational data science. Especially, if no approach is taken to rate whether a data universe discovery should be useful or not. We suggest that rigorous approaches to measure the usefulness of data universe discoveries might mitigate such a divergence.

Caroline Mazini Rodrigues, Nicolas Boutry, Laurent Najman

Challenges persist in providing interpretable explanations for neural network reasoning in explainable AI (xAI). Existing methods like Integrated Gradients produce noisy maps, and LIME, while intuitive, may deviate from the model's reasoning. We introduce a framework that uses hierarchical segmentation techniques for faithful and interpretable explanations of Convolutional Neural Networks (CNNs). Our method constructs model-based hierarchical segmentations that maintain the model's reasoning fidelity and allows both human-centric and model-centric segmentation. This approach offers multiscale explanations, aiding bias identification and enhancing understanding of neural network decision-making. Experiments show that our framework, xAiTrees, delivers highly interpretable and faithful model explanations, not only surpassing traditional xAI methods but shedding new light on a novel approach to enhancing xAI interpretability.

Aditya Challa, Sravan Danda, Laurent Najman et al.

Standard ML models fail to infer the context distribution and suitably adapt. For instance, the learning fails when the underlying distribution is actually a mixture of distributions with contradictory labels. Learning also fails if there is a shift between train and test distributions. Standard neural network architectures like MLPs or CNNs are not equipped to handle this. In this article, we propose a simple activation function, quantile activation (QAct), that addresses this problem without significantly increasing computational costs. The core idea is to "adapt" the outputs of each neuron to its context distribution. The proposed quantile activation (QAct) outputs the relative quantile position of neuron activations within their context distribution, diverging from the direct numerical outputs common in traditional networks. A specific case of the above failure mode is when there is an inherent distribution shift, i.e the test distribution differs slightly from the train distribution. We validate the proposed activation function under covariate shifts, using datasets designed to test robustness against distortions. Our results demonstrate significantly better generalization across distortions compared to conventional classifiers and other adaptive methods, across various architectures. Although this paper presents a proof of concept, we find that this approach unexpectedly outperforms DINOv2 (small), despite DINOv2 being trained with a much larger network and dataset.

Caroline Mazini Rodrigues, Nicolas Boutry, Laurent Najman

The explication of Convolutional Neural Networks (CNN) through xAI techniques often poses challenges in interpretation. The inherent complexity of input features, notably pixels extracted from images, engenders complex correlations. Gradient-based methodologies, exemplified by Integrated Gradients (IG), effectively demonstrate the significance of these features. Nevertheless, the conversion of these explanations into images frequently yields considerable noise. Presently, we introduce GAD (Gradient Artificial Distancing) as a supportive framework for gradient-based techniques. Its primary objective is to accentuate influential regions by establishing distinctions between classes. The essence of GAD is to limit the scope of analysis during visualization and, consequently reduce image noise. Empirical investigations involving occluded images have demonstrated that the identified regions through this methodology indeed play a pivotal role in facilitating class differentiation.

Yuliang Gu, Zhichao Sun, Tian Chen et al.

Semi-supervised image segmentation has attracted great attention recently. The key is how to leverage unlabeled images in the training process. Most methods maintain consistent predictions of the unlabeled images under variations (e.g., adding noise/perturbations, or creating alternative versions) in the image and/or model level. In most image-level variation, medical images often have prior structure information, which has not been well explored. In this paper, we propose novel dual structure-aware image filterings (DSAIF) as the image-level variations for semi-supervised medical image segmentation. Motivated by connected filtering that simplifies image via filtering in structure-aware tree-based image representation, we resort to the dual contrast invariant Max-tree and Min-tree representation. Specifically, we propose a novel connected filtering that removes topologically equivalent nodes (i.e. connected components) having no siblings in the Max/Min-tree. This results in two filtered images preserving topologically critical structure. Applying the proposed DSAIF to mutually supervised networks decreases the consensus of their erroneous predictions on unlabeled images. This helps to alleviate the confirmation bias issue of overfitting to noisy pseudo labels of unlabeled images, and thus effectively improves the segmentation performance. Extensive experimental results on three benchmark datasets demonstrate that the proposed method significantly/consistently outperforms some state-of-the-art methods. The source codes will be publicly available.

Ketan Bacchuwar, Jean Cousty, Régis Vaillant et al.

In this article, we present the work towards improving the overall workflow of the Percutaneous Coronary Interventions (PCI) procedures by capacitating the imaging instruments to precisely monitor the steps of the procedure. In the long term, such capabilities can be used to optimize the image acquisition to reduce the amount of dose or contrast media employed during the procedure. We present the automatic VOIDD algorithm to detect the vessel of intervention which is going to be treated during the procedure by combining information from the vessel image with contrast agent injection and images acquired during guidewire tip navigation. Due to the robust guidewire tip segmentation method, this algorithm is also able to automatically detect the sequence corresponding to guidewire navigation. We present an evaluation methodology which characterizes the correctness of the guide wire tip detection and correct identification of the vessel navigated during the procedure. On a dataset of 2213 images from 8 sequences of 4 patients, VOIDD identifies vessel-of-intervention with accuracy in the range of 88% or above and absence of tip with accuracy in range of 98% or above depending on the test case.

Yongchao Xu, Thierry Géraud, Laurent Najman

Hierarchies, such as the tree of shapes, are popular representations for image simplification and segmentation thanks to their multiscale structures. Selecting meaningful level lines (boundaries of shapes) yields to simplify image while preserving intact salient structures. Many image simplification and segmentation methods are driven by the optimization of an energy functional, for instance the celebrated Mumford-Shah functional. In this paper, we propose an efficient approach to hierarchical image simplification and segmentation based on the minimization of the piecewise-constant Mumford-Shah functional. This method conforms to the current trend that consists in producing hierarchical results rather than a unique partition. Contrary to classical approaches which compute optimal hierarchical segmentations from an input hierarchy of segmentations, we rely on the tree of shapes, a unique and well-defined representation equivalent to the image. Simply put, we compute for each level line of the image an attribute function that characterizes its persistence under the energy minimization. Then we stack the level lines from meaningless ones to salient ones through a saliency map based on extinction values defined on the tree-based shape space. Qualitative illustrations and quantitative evaluation on Weizmann segmentation evaluation database demonstrate the state-of-the-art performance of our method.

Jean Cousty, Laurent Najman, Yukiko Kenmochi et al.

We study three representations of hierarchies of partitions: dendrograms (direct representations), saliency maps, and minimum spanning trees. We provide a new bijection between saliency maps and hierarchies based on quasi-flat zones as used in image processing and characterize saliency maps and minimum spanning trees as solutions to constrained minimization problems where the constraint is quasi-flat zones preservation. In practice, these results form a toolkit for new hierarchical methods where one can choose the most convenient representation. They also invite us to process non-image data with morphological hierarchies.

Laurent Najman, Jean Cousty

This survey paper aims at providing a "literary" anthology of mathematical morphology on graphs. It describes in the English language many ideas stemming from a large number of different papers, hence providing a unified view of an active and diverse field of research.

Camille Couprie, Clément Farabet, Laurent Najman et al.

This work addresses multi-class segmentation of indoor scenes with RGB-D inputs. While this area of research has gained much attention recently, most works still rely on hand-crafted features. In contrast, we apply a multiscale convolutional network to learn features directly from the images and the depth information. We obtain state-of-the-art on the NYU-v2 depth dataset with an accuracy of 64.5%. We illustrate the labeling of indoor scenes in videos sequences that could be processed in real-time using appropriate hardware such as an FPGA.

Roland Levillain, Thierry Géraud, Laurent Najman

Digital Geometry software should reflect the generality of the underlying mathe- matics: mapping the latter to the former requires genericity. By designing generic solutions, one can effectively reuse digital geometry data structures and algorithms. We propose an image processing framework focused on the Generic Programming paradigm in which an algorithm on the paper can be turned into a single code, written once and usable with various input types. This approach enables users to design and implement new methods at a lower cost, try cross-domain experiments and help generalize results

Silvio Jamil F. Guimarães, Jean Cousty, Yukiko Kenmochi et al.

Hierarchical image segmentation provides region-oriented scalespace, i.e., a set of image segmentations at different detail levels in which the segmentations at finer levels are nested with respect to those at coarser levels. Most image segmentation algorithms, such as region merging algorithms, rely on a criterion for merging that does not lead to a hierarchy, and for which the tuning of the parameters can be difficult. In this work, we propose a hierarchical graph based image segmentation relying on a criterion popularized by Felzenzwalb and Huttenlocher. We illustrate with both real and synthetic images, showing efficiency, ease of use, and robustness of our method.

Yongchao Xu, Thierry Géraud, Laurent Najman

Connected operators are filtering tools that act by merging elementary regions of an image. A popular strategy is based on tree-based image representations: for example, one can compute an attribute on each node of the tree and keep only the nodes for which the attribute is sufficiently strong. This operation can be seen as a thresholding of the tree, seen as a graph whose nodes are weighted by the attribute. Rather than being satisfied with a mere thresholding, we propose to expand on this idea, and to apply connected filters on this latest graph. Consequently, the filtering is done not in the space of the image, but on the space of shapes build from the image. Such a processing is a generalization of the existing tree-based connected operators. Indeed, the framework includes classical existing connected operators by attributes. It also allows us to propose a class of novel connected operators from the leveling family, based on shape attributes. Finally, we also propose a novel class of self-dual connected operators that we call morphological shapings.

Clément Farabet, Camille Couprie, Laurent Najman et al.

Scene parsing, or semantic segmentation, consists in labeling each pixel in an image with the category of the object it belongs to. It is a challenging task that involves the simultaneous detection, segmentation and recognition of all the objects in the image. The scene parsing method proposed here starts by computing a tree of segments from a graph of pixel dissimilarities. Simultaneously, a set of dense feature vectors is computed which encodes regions of multiple sizes centered on each pixel. The feature extractor is a multiscale convolutional network trained from raw pixels. The feature vectors associated with the segments covered by each node in the tree are aggregated and fed to a classifier which produces an estimate of the distribution of object categories contained in the segment. A subset of tree nodes that cover the image are then selected so as to maximize the average "purity" of the class distributions, hence maximizing the overall likelihood that each segment will contain a single object. The convolutional network feature extractor is trained end-to-end from raw pixels, alleviating the need for engineered features. After training, the system is parameter free. The system yields record accuracies on the Stanford Background Dataset (8 classes), the Sift Flow Dataset (33 classes) and the Barcelona Dataset (170 classes) while being an order of magnitude faster than competing approaches, producing a 320 \times 240 image labeling in less than 1 second.