Aleix M. Martı́nez

Samarth Gupta, Raghudeep Gadde, Rui Chen et al.

We challenge a fundamental assumption of diffusion models, namely, that a large number of latent-states or time-steps is required for training so that the reverse generative process is close to a Gaussian. We first show that with careful selection of a noise schedule, diffusion models trained over a small number of latent states (i.e. $T \sim 32$) match the performance of models trained over a much large number of latent states ($T \sim 1,000$). Second, we push this limit (on the minimum number of latent states required) to a single latent-state, which we refer to as complete disentanglement in T-space. We show that high quality samples can be easily generated by the disentangled model obtained by combining several independently trained single latent-state models. We provide extensive experiments to show that the proposed disentangled model provides 4-6$\times$ faster convergence measured across a variety of metrics on two different datasets.

Giorgio Giannone, Ruoteng Li, Qianli Feng et al.

Vision-language models (VLMs) have demonstrated remarkable potential in integrating visual and linguistic information, but their performance is often constrained by the need for extensive, high-quality image-text training data. Curation of these image-text pairs is both time-consuming and computationally expensive. To address this challenge, we introduce SVP (Sampling-based Visual Projection), a novel framework that enhances vision-language alignment without relying on manually curated text-image pairs or preference annotation. SVP leverages a small set of manually selected images, self-captioning and a pre-trained grounding model as a feedback mechanism to elicit latent information in VLMs. We evaluate our approach across six key areas: captioning, referring, visual question answering, multitasking, hallucination control, and object recall. Results demonstrate significant improvements, including a 14 % average improvement in captioning tasks, up to 12 % increase in object recall, and significantly reduced hallucinations, while maintaining question-answering capabilities. Using SVP, a small VLM achieves hallucination reductions similar to a model five times larger, while a VLM with initially poor referring capabilities more than doubles its performance, approaching parity with a model twice its size.

Qianli Feng, Viraj Shah, Raghudeep Gadde et al.

To edit a real photo using Generative Adversarial Networks (GANs), we need a GAN inversion algorithm to identify the latent vector that perfectly reproduces it. Unfortunately, whereas existing inversion algorithms can synthesize images similar to real photos, they cannot generate the identical clones needed in most applications. Here, we derive an algorithm that achieves near perfect reconstructions of photos. Rather than relying on encoder- or optimization-based methods to find an inverse mapping on a fixed generator $G(\cdot)$, we derive an approach to locally adjust $G(\cdot)$ to more optimally represent the photos we wish to synthesize. This is done by locally tweaking the learned mapping $G(\cdot)$ s.t. $\| {\bf x} - G({\bf z}) \|<ε$, with ${\bf x}$ the photo we wish to reproduce, ${\bf z}$ the latent vector, $\|\cdot\|$ an appropriate metric, and $ε> 0$ a small scalar. We show that this approach can not only produce synthetic images that are indistinguishable from the real photos we wish to replicate, but that these images are readily editable. We demonstrate the effectiveness of the derived algorithm on a variety of datasets including human faces, animals, and cars, and discuss its importance for diversity and inclusion.

Qianli Feng, Chenqi Guo, Fabian Benitez-Quiroz et al.

Do GANs replicate training images? Previous studies have shown that GANs do not seem to replicate training data without significant change in the training procedure. This leads to a series of research on the exact condition needed for GANs to overfit to the training data. Although a number of factors has been theoretically or empirically identified, the effect of dataset size and complexity on GANs replication is still unknown. With empirical evidence from BigGAN and StyleGAN2, on datasets CelebA, Flower and LSUN-bedroom, we show that dataset size and its complexity play an important role in GANs replication and perceptual quality of the generated images. We further quantify this relationship, discovering that replication percentage decays exponentially with respect to dataset size and complexity, with a shared decaying factor across GAN-dataset combinations. Meanwhile, the perceptual image quality follows a U-shape trend w.r.t dataset size. This finding leads to a practical tool for one-shot estimation on minimal dataset size to prevent GAN replication which can be used to guide datasets construction and selection.

Guha Balakrishnan, Raghudeep Gadde, Aleix Martinez et al.

We present a method for finding paths in a deep generative model's latent space that can maximally vary one set of image features while holding others constant. Crucially, unlike past traversal approaches, ours can manipulate multidimensional features of an image such as facial identity and pixels within a specified region. Our method is principled and conceptually simple: optimal traversal directions are chosen by maximizing differential changes to one feature set such that changes to another set are negligible. We show that this problem is nearly equivalent to one of Rayleigh quotient maximization, and provide a closed-form solution to it based on solving a generalized eigenvalue equation. We use repeated computations of the corresponding optimal directions, which we call Rayleigh EigenDirections (REDs), to generate appropriately curved paths in latent space. We empirically evaluate our method using StyleGAN2 on two image domains: faces and living rooms. We show that our method is capable of controlling various multidimensional features out of the scope of previous latent space traversal methods: face identity, spatial frequency bands, pixels within a region, and the appearance and position of an object. Our work suggests that a wealth of opportunities lies in the local analysis of the geometry and semantics of latent spaces.

Stuart Synakowski, Fabian Benitez-Quiroz, Aleix M. Martinez

Recently, methods have been developed to accurately predict the testing performance of a Deep Neural Network (DNN) on a particular task, given statistics of its underlying topological structure. However, further leveraging this newly found insight for practical applications is intractable due to the high computational cost in terms of time and memory. In this work, we define a new class of topological features that accurately characterize the progress of learning while being quick to compute during running time. Additionally, our proposed topological features are readily equipped for backpropagation, meaning that they can be incorporated in end-to-end training. Our newly developed practical topological characterization of DNNs allows for an additional set of applications. We first show we can predict the performance of a DNN without a testing set and without the need for high-performance computing. We also demonstrate our topological characterization of DNNs is effective in estimating task similarity. Lastly, we show we can induce learning in DNNs by actively constraining the DNN's topological structure. This opens up new avenues in constricting the underlying structure of DNNs in a meta-learning framework.

Jeffrey Wen, Fabian Benitez-Quiroz, Qianli Feng et al.

In just a few years, the photo-realism of images synthesized by Generative Adversarial Networks (GANs) has gone from somewhat reasonable to almost perfect largely by increasing the complexity of the networks, e.g., adding layers, intermediate latent spaces, style-transfer parameters, etc. This trajectory has led many of the state-of-the-art GANs to be inaccessibly large, disengaging many without large computational resources. Recognizing this, we explore a method for squeezing additional performance from existing, low-complexity GANs. Formally, we present an unsupervised method to find a direction in the latent space that aligns with improved photo-realism. Our approach leaves the network unchanged while enhancing the fidelity of the generated image. We use a simple generator inversion to find the direction in the latent space that results in the smallest change in the image space. Leveraging the learned structure of the latent space, we find moving in this direction corrects many image artifacts and brings the image into greater realism. We verify our findings qualitatively and quantitatively, showing an improvement in Frechet Inception Distance (FID) exists along our trajectory which surpasses the original GAN and other approaches including a supervised method. We expand further and provide an optimization method to automatically select latent vectors along the path that balance the variation and realism of samples. We apply our method to several diverse datasets and three architectures of varying complexity to illustrate the generalizability of our approach. By expanding the utility of low-complexity and existing networks, we hope to encourage the democratization of GANs.

Stuart Synakowski, Qianli Feng, Aleix Martinez

Computer vision algorithms performance are near or superior to humans in the visual problems including object recognition (especially those of fine-grained categories), segmentation, and 3D object reconstruction from 2D views. Humans are, however, capable of higher-level image analyses. A clear example, involving theory of mind, is our ability to determine whether a perceived behavior or action was performed intentionally or not. In this paper, we derive an algorithm that can infer whether the behavior of an agent in a scene is intentional or unintentional based on its 3D kinematics, using the knowledge of self-propelled motion, Newtonian motion and their relationship. We show how the addition of this basic knowledge leads to a simple, unsupervised algorithm. To test the derived algorithm, we constructed three dedicated datasets from abstract geometric animation to realistic videos of agents performing intentional and non-intentional actions. Experiments on these datasets show that our algorithm can recognize whether an action is intentional or not, even without training data. The performance is comparable to various supervised baselines quantitatively, with sensible intentionality segmentation qualitatively.

Samuel Rivera, Catherine A. Best, Hyungwook Yim et al.

Visual categorization and learning of visual categories exhibit early onset, however the underlying mechanisms of early categorization are not well understood. The main limiting factor for examining these mechanisms is the limited duration of infant cooperation (10-15 minutes), which leaves little room for multiple test trials. With its tight link to visual attention, eye tracking is a promising method for getting access to the mechanisms of category learning. But how should researchers decide which aspects of the rich eye tracking data to focus on? To date, eye tracking variables are generally handpicked, which may lead to biases in the eye tracking data. Here, we propose an automated method for selecting eye tracking variables based on analyses of their usefulness to discriminate learners from non-learners of visual categories. We presented infants and adults with a category learning task and tracked their eye movements. We then extracted an over-complete set of eye tracking variables encompassing durations, probabilities, latencies, and the order of fixations and saccadic eye movements. We compared three statistical techniques for identifying those variables among this large set that are useful for discriminating learners form non-learners: ANOVA ranking, Bayes ranking, and L1 regularized logistic regression. We found remarkable agreement between these methods in identifying a small set of discriminant variables. Moreover, the same eye tracking variables allow us to classify category learners from non-learners among adults and 6- to 8-month-old infants with accuracies above 71%.

Ciprian Corneanu, Meysam Madadi, Sergio Escalera et al.

Deep Neural Networks (DNNs) have revolutionized computer vision. We now have DNNs that achieve top (performance) results in many problems, including object recognition, facial expression analysis, and semantic segmentation, to name but a few. The design of the DNNs that achieve top results is, however, non-trivial and mostly done by trail-and-error. That is, typically, researchers will derive many DNN architectures (i.e., topologies) and then test them on multiple datasets. However, there are no guarantees that the selected DNN will perform well in the real world. One can use a testing set to estimate the performance gap between the training and testing sets, but avoiding overfitting-to-the-testing-data is almost impossible. Using a sequestered testing dataset may address this problem, but this requires a constant update of the dataset, a very expensive venture. Here, we derive an algorithm to estimate the performance gap between training and testing that does not require any testing dataset. Specifically, we derive a number of persistent topology measures that identify when a DNN is learning to generalize to unseen samples. This allows us to compute the DNN's testing error on unseen samples, even when we do not have access to them. We provide extensive experimental validation on multiple networks and datasets to demonstrate the feasibility of the proposed approach.

Ramprakash Srinivasan, Aleix M. Martinez

Automatic recognition of emotion from facial expressions is an intense area of research, with a potentially long list of important application. Yet, the study of emotion requires knowing which facial expressions are used within and across cultures in the wild, not in controlled lab conditions; but such studies do not exist. Which and how many cross-cultural and cultural-specific facial expressions do people commonly use? And, what affect variables does each expression communicate to observers? If we are to design technology that understands the emotion of users, we need answers to these two fundamental questions. In this paper, we present the first large-scale study of the production and visual perception of facial expressions of emotion in the wild. We find that of the 16,384 possible facial configurations that people can theoretically produce, only 35 are successfully used to transmit emotive information across cultures, and only 8 within a smaller number of cultures. Crucially, we find that visual analysis of cross-cultural expressions yields consistent perception of emotion categories and valence, but not arousal. In contrast, visual analysis of cultural-specific expressions yields consistent perception of valence and arousal, but not of emotion categories. Additionally, we find that the number of expressions used to communicate each emotion is also different, e.g., 17 expressions transmit happiness, but only 1 is used to convey disgust.

Albert Pumarola, Antonio Agudo, Aleix M. Martinez et al.

Recent advances in Generative Adversarial Networks (GANs) have shown impressive results for task of facial expression synthesis. The most successful architecture is StarGAN, that conditions GANs generation process with images of a specific domain, namely a set of images of persons sharing the same expression. While effective, this approach can only generate a discrete number of expressions, determined by the content of the dataset. To address this limitation, in this paper, we introduce a novel GAN conditioning scheme based on Action Units (AU) annotations, which describes in a continuous manifold the anatomical facial movements defining a human expression. Our approach allows controlling the magnitude of activation of each AU and combine several of them. Additionally, we propose a fully unsupervised strategy to train the model, that only requires images annotated with their activated AUs, and exploit attention mechanisms that make our network robust to changing backgrounds and lighting conditions. Extensive evaluation show that our approach goes beyond competing conditional generators both in the capability to synthesize a much wider range of expressions ruled by anatomically feasible muscle movements, as in the capacity of dealing with images in the wild.

C. Fabian Benitez-Quiroz, Ramprakash Srinivasan, Qianli Feng et al.

This paper details the methodology and results of the EmotioNet challenge. This challenge is the first to test the ability of computer vision algorithms in the automatic analysis of a large number of images of facial expressions of emotion in the wild. The challenge was divided into two tracks. The first track tested the ability of current computer vision algorithms in the automatic detection of action units (AUs). Specifically, we tested the detection of 11 AUs. The second track tested the algorithms' ability to recognize emotion categories in images of facial expressions. Specifically, we tested the recognition of 16 basic and compound emotion categories. The results of the challenge suggest that current computer vision and machine learning algorithms are unable to reliably solve these two tasks. The limitations of current algorithms are more apparent when trying to recognize emotion. We also show that current algorithms are not affected by mild resolution changes, small occluders, gender or age, but that 3D pose is a major limiting factor on performance. We provide an in-depth discussion of the points that need special attention moving forward.

Ruiqi Zhao, Yan Wang, Aleix Martinez

Three-dimensional shape reconstruction of 2D landmark points on a single image is a hallmark of human vision, but is a task that has been proven difficult for computer vision algorithms. We define a feed-forward deep neural network algorithm that can reconstruct 3D shapes from 2D landmark points almost perfectly (i.e., with extremely small reconstruction errors), even when these 2D landmarks are from a single image. Our experimental results show an improvement of up to two-fold over state-of-the-art computer vision algorithms; 3D shape reconstruction of human faces is given at a reconstruction error < .004, cars at .0022, human bodies at .022, and highly-deformable flags at an error of .0004. Our algorithm was also a top performer at the 2016 3D Face Alignment in the Wild Challenge competition (done in conjunction with the European Conference on Computer Vision, ECCV) that required the reconstruction of 3D face shape from a single image. The derived algorithm can be trained in a couple hours and testing runs at more than 1, 000 frames/s on an i7 desktop. We also present an innovative data augmentation approach that allows us to train the system efficiently with small number of samples. And the system is robust to noise (e.g., imprecise landmark points) and missing data (e.g., occluded or undetected landmark points).