Pietro Perona

Justin Kay, Peter Kulits, Suzanne Stathatos et al. · mit

We present the Caltech Fish Counting Dataset (CFC), a large-scale dataset for detecting, tracking, and counting fish in sonar videos. We identify sonar videos as a rich source of data for advancing low signal-to-noise computer vision applications and tackling domain generalization in multiple-object tracking (MOT) and counting. In comparison to existing MOT and counting datasets, which are largely restricted to videos of people and vehicles in cities, CFC is sourced from a natural-world domain where targets are not easily resolvable and appearance features cannot be easily leveraged for target re-identification. With over half a million annotations in over 1,500 videos sourced from seven different sonar cameras, CFC allows researchers to train MOT and counting algorithms and evaluate generalization performance at unseen test locations. We perform extensive baseline experiments and identify key challenges and opportunities for advancing the state of the art in generalization in MOT and counting.

Jennifer J. Sun, Lili Karashchuk, Amil Dravid et al. · cambridge, uw

Quantifying motion in 3D is important for studying the behavior of humans and other animals, but manual pose annotations are expensive and time-consuming to obtain. Self-supervised keypoint discovery is a promising strategy for estimating 3D poses without annotations. However, current keypoint discovery approaches commonly process single 2D views and do not operate in the 3D space. We propose a new method to perform self-supervised keypoint discovery in 3D from multi-view videos of behaving agents, without any keypoint or bounding box supervision in 2D or 3D. Our method, BKinD-3D, uses an encoder-decoder architecture with a 3D volumetric heatmap, trained to reconstruct spatiotemporal differences across multiple views, in addition to joint length constraints on a learned 3D skeleton of the subject. In this way, we discover keypoints without requiring manual supervision in videos of humans and rats, demonstrating the potential of 3D keypoint discovery for studying behavior.

Hao Liang, Pietro Perona, Guha Balakrishnan

We propose an experimental method for measuring bias in face recognition systems. Existing methods to measure bias depend on benchmark datasets that are collected in the wild and annotated for protected (e.g., race, gender) and non-protected (e.g., pose, lighting) attributes. Such observational datasets only permit correlational conclusions, e.g., "Algorithm A's accuracy is different on female and male faces in dataset X.". By contrast, experimental methods manipulate attributes individually and thus permit causal conclusions, e.g., "Algorithm A's accuracy is affected by gender and skin color." Our method is based on generating synthetic faces using a neural face generator, where each attribute of interest is modified independently while leaving all other attributes constant. Human observers crucially provide the ground truth on perceptual identity similarity between synthetic image pairs. We validate our method quantitatively by evaluating race and gender biases of three research-grade face recognition models. Our synthetic pipeline reveals that for these algorithms, accuracy is lower for Black and East Asian population subgroups. Our method can also quantify how perceptual changes in attributes affect face identity distances reported by these models. Our large synthetic dataset, consisting of 48,000 synthetic face image pairs (10,200 unique synthetic faces) and 555,000 human annotations (individual attributes and pairwise identity comparisons) is available to researchers in this important area.

Elijah Cole, Kimberly Wilber, Grant Van Horn et al.

Weakly supervised object localization (WSOL) aims to learn representations that encode object location using only image-level category labels. However, many objects can be labeled at different levels of granularity. Is it an animal, a bird, or a great horned owl? Which image-level labels should we use? In this paper we study the role of label granularity in WSOL. To facilitate this investigation we introduce iNatLoc500, a new large-scale fine-grained benchmark dataset for WSOL. Surprisingly, we find that choosing the right training label granularity provides a much larger performance boost than choosing the best WSOL algorithm. We also show that changing the label granularity can significantly improve data efficiency.

Neehar Kondapaneni, Markus Marks, Manuel Knott et al.

Diffusion models are generative models with impressive text-to-image synthesis capabilities and have spurred a new wave of creative methods for classical machine learning tasks. However, the best way to harness the perceptual knowledge of these generative models for visual tasks is still an open question. Specifically, it is unclear how to use the prompting interface when applying diffusion backbones to vision tasks. We find that automatically generated captions can improve text-image alignment and significantly enhance a model's cross-attention maps, leading to better perceptual performance. Our approach improves upon the current state-of-the-art (SOTA) in diffusion-based semantic segmentation on ADE20K and the current overall SOTA for depth estimation on NYUv2. Furthermore, our method generalizes to the cross-domain setting. We use model personalization and caption modifications to align our model to the target domain and find improvements over unaligned baselines. Our cross-domain object detection model, trained on Pascal VOC, achieves SOTA results on Watercolor2K. Our cross-domain segmentation method, trained on Cityscapes, achieves SOTA results on Dark Zurich-val and Nighttime Driving. Project page: https://www.vision.caltech.edu/tadp/. Code: https://github.com/damaggu/TADP.

Elijah Cole, Grant Van Horn, Christian Lange et al.

Estimating the geographical range of a species from sparse observations is a challenging and important geospatial prediction problem. Given a set of locations where a species has been observed, the goal is to build a model to predict whether the species is present or absent at any location. This problem has a long history in ecology, but traditional methods struggle to take advantage of emerging large-scale crowdsourced datasets which can include tens of millions of records for hundreds of thousands of species. In this work, we use Spatial Implicit Neural Representations (SINRs) to jointly estimate the geographical range of 47k species simultaneously. We find that our approach scales gracefully, making increasingly better predictions as we increase the number of species and the amount of data per species when training. To make this problem accessible to machine learning researchers, we provide four new benchmarks that measure different aspects of species range estimation and spatial representation learning. Using these benchmarks, we demonstrate that noisy and biased crowdsourced data can be combined with implicit neural representations to approximate expert-developed range maps for many species.

Markus Marks, Manuel Knott, Neehar Kondapaneni et al.

Self-supervised learning (SSL) is a machine learning approach where the data itself provides supervision, eliminating the need for external labels. The model is forced to learn about the data structure or context by solving a pretext task. With SSL, models can learn from abundant and cheap unlabeled data, significantly reducing the cost of training models where labels are expensive or inaccessible. In Computer Vision, SSL is widely used as pre-training followed by a downstream task, such as supervised transfer, few-shot learning on smaller labeled data sets, and/or unsupervised clustering. Unfortunately, it is infeasible to evaluate SSL methods on all possible downstream tasks and objectively measure the quality of the learned representation. Instead, SSL methods are evaluated using in-domain evaluation protocols, such as fine-tuning, linear probing, and k-nearest neighbors (kNN). However, it is not well understood how well these evaluation protocols estimate the representation quality of a pre-trained model for different downstream tasks under different conditions, such as dataset, metric, and model architecture. We study how classification-based evaluation protocols for SSL correlate and how well they predict downstream performance on different dataset types. Our study includes eleven common image datasets and 26 models that were pre-trained with different SSL methods or have different model backbones. We find that in-domain linear/kNN probing protocols are, on average, the best general predictors for out-of-domain performance. We further investigate the importance of batch normalization and evaluate how robust correlations are for different kinds of dataset domain shifts. We challenge assumptions about the relationship between discriminative and generative self-supervised methods, finding that most of their performance differences can be explained by changes to model backbones.

Daniel Khalil, Christina Liu, Pietro Perona et al.

The study of social interactions and collective behaviors through multi-agent video analysis is crucial in biology. While self-supervised keypoint discovery has emerged as a promising solution to reduce the need for manual keypoint annotations, existing methods often struggle with videos containing multiple interacting agents, especially those of the same species and color. To address this, we introduce B-KinD-multi, a novel approach that leverages pre-trained video segmentation models to guide keypoint discovery in multi-agent scenarios. This eliminates the need for time-consuming manual annotations on new experimental settings and organisms. Extensive evaluations demonstrate improved keypoint regression and downstream behavioral classification in videos of flies, mice, and rats. Furthermore, our method generalizes well to other species, including ants, bees, and humans, highlighting its potential for broad applications in automated keypoint annotation for multi-agent behavior analysis. Code available under: https://danielpkhalil.github.io/B-KinD-Multi

Carina I. Hausladen, Manuel Knott, Colin F. Camerer et al.

We explore social perception of human faces in CLIP, a widely used open-source vision-language model. To this end, we compare the similarity in CLIP embeddings between different textual prompts and a set of face images. Our textual prompts are constructed from well-validated social psychology terms denoting social perception. The face images are synthetic and are systematically and independently varied along six dimensions: the legally protected attributes of age, gender, and race, as well as facial expression, lighting, and pose. Independently and systematically manipulating face attributes allows us to study the effect of each on social perception and avoids confounds that can occur in wild-collected data due to uncontrolled systematic correlations between attributes. Thus, our findings are experimental rather than observational. Our main findings are three. First, while CLIP is trained on the widest variety of images and texts, it is able to make fine-grained human-like social judgments on face images. Second, age, gender, and race do systematically impact CLIP's social perception of faces, suggesting an undesirable bias in CLIP vis-a-vis legally protected attributes. Most strikingly, we find a strong pattern of bias concerning the faces of Black women, where CLIP produces extreme values of social perception across different ages and facial expressions. Third, facial expression impacts social perception more than age and lighting as much as age. The last finding predicts that studies that do not control for unprotected visual attributes may reach the wrong conclusions on bias. Our novel method of investigation, which is founded on the social psychology literature and on the experiments involving the manipulation of individual attributes, yields sharper and more reliable observations than previous observational methods and may be applied to study biases in any vision-language model.

Riccardo Fogliato, Pratik Patil, Pietro Perona

Matching algorithms are commonly used to predict matches between items in a collection. For example, in 1:1 face verification, a matching algorithm predicts whether two face images depict the same person. Accurately assessing the uncertainty of the error rates of such algorithms can be challenging when data are dependent and error rates are low, two aspects that have been often overlooked in the literature. In this work, we review methods for constructing confidence intervals for error rates in 1:1 matching tasks. We derive and examine the statistical properties of these methods, demonstrating how coverage and interval width vary with sample size, error rates, and degree of data dependence on both analysis and experiments with synthetic and real-world datasets. Based on our findings, we provide recommendations for best practices for constructing confidence intervals for error rates in 1:1 matching tasks.

Neehar Kondapaneni, Pietro Perona, Oisin Mac Aodha

Each year, thousands of people learn new visual categorization tasks -- radiologists learn to recognize tumors, birdwatchers learn to distinguish similar species, and crowd workers learn how to annotate valuable data for applications like autonomous driving. As humans learn, their brain updates the visual features it extracts and attend to, which ultimately informs their final classification decisions. In this work, we propose a novel task of tracing the evolving classification behavior of human learners as they engage in challenging visual classification tasks. We propose models that jointly extract the visual features used by learners as well as predicting the classification functions they utilize. We collect three challenging new datasets from real human learners in order to evaluate the performance of different visual knowledge tracing methods. Our results show that our recurrent models are able to predict the classification behavior of human learners on three challenging medical image and species identification tasks.

Jennifer J. Sun, Markus Marks, Andrew Ulmer et al.

We introduce MABe22, a large-scale, multi-agent video and trajectory benchmark to assess the quality of learned behavior representations. This dataset is collected from a variety of biology experiments, and includes triplets of interacting mice (4.7 million frames video+pose tracking data, 10 million frames pose only), symbiotic beetle-ant interactions (10 million frames video data), and groups of interacting flies (4.4 million frames of pose tracking data). Accompanying these data, we introduce a panel of real-life downstream analysis tasks to assess the quality of learned representations by evaluating how well they preserve information about the experimental conditions (e.g. strain, time of day, optogenetic stimulation) and animal behavior. We test multiple state-of-the-art self-supervised video and trajectory representation learning methods to demonstrate the use of our benchmark, revealing that methods developed using human action datasets do not fully translate to animal datasets. We hope that our benchmark and dataset encourage a broader exploration of behavior representation learning methods across species and settings.

Sevan Brodjian, Michael Hobley, Pietro Perona

Sonar is often the only modality suitable for high-resolution imaging underwater due to light attenuation and turbidity. Forward-looking imaging sonar provides measurements over range and horizontal angle but collapses vertical structure into a flat image, creating ambiguities that make 3D recovery challenging. A common use case for imaging sonar is underwater terrain mapping (bathymetry), yet current methods require many views, expensive multi-sensor setups, or significant training data, which limits use and adaptability to new environments. We present a training-free method that recovers bathymetry from a single sonar image in under 30 seconds via differentiable rendering, conditioned on a known seafloor tilt. To our knowledge, this is the first differentiable rendering approach for single-view height recovery in sonar. Our method implements differentiable sonar ray tracing and optimizes an explicit height field to reproduce the target image. On synthetic datasets, our approach outperforms a supervised CNN under distribution shift and remains close on rough terrain, while the CNN wins in-distribution. By modeling physically grounded priors of the sonar process, our method adapts across sensor configurations and environments without training data.

Justin Kay, Timm Haucke, Suzanne Stathatos et al. · mit

Object detectors often perform poorly on data that differs from their training set. Domain adaptive object detection (DAOD) methods have recently demonstrated strong results on addressing this challenge. Unfortunately, we identify systemic benchmarking pitfalls that call past results into question and hamper further progress: (a) Overestimation of performance due to underpowered baselines, (b) Inconsistent implementation practices preventing transparent comparisons of methods, and (c) Lack of generality due to outdated backbones and lack of diversity in benchmarks. We address these problems by introducing: (1) A unified benchmarking and implementation framework, Align and Distill (ALDI), enabling comparison of DAOD methods and supporting future development, (2) A fair and modern training and evaluation protocol for DAOD that addresses benchmarking pitfalls, (3) A new DAOD benchmark dataset, CFC-DAOD, enabling evaluation on diverse real-world data, and (4) A new method, ALDI++, that achieves state-of-the-art results by a large margin. ALDI++ outperforms the previous state-of-the-art by +3.5 AP50 on Cityscapes to Foggy Cityscapes, +5.7 AP50 on Sim10k to Cityscapes (where ours is the only method to outperform a fair baseline), and +0.6 AP50 on CFC Kenai to Channel. ALDI and ALDI++ are architecture-agnostic, setting a new state-of-the-art for YOLO and DETR-based DAOD as well without additional hyperparameter tuning. Our framework, dataset, and state-of-the-art method offer a critical reset for DAOD and provide a strong foundation for future research. Code and data are available: https://github.com/justinkay/aldi and https://github.com/visipedia/caltech-fish-counting.

Raphi Kang, Yue Song, Georgia Gkioxari et al.

Contrastive Language-Image Pre-Training (CLIP) is a popular method for learning multimodal latent spaces with well-organized semantics. Despite its wide range of applications, CLIP's latent space is known to fail at handling complex visual-textual interactions. Recent works attempt to address its shortcomings with data-centric or algorithmic approaches. But what if the problem is more fundamental, and lies in the geometry of CLIP? Toward this end, we rigorously analyze CLIP's latent space properties, and prove that no CLIP-like joint embedding space exists which can correctly do any two of the following at the same time: 1. represent basic descriptions and image content, 2. represent attribute binding, 3. represent spatial location and relationships, 4. represent negation. Informed by this analysis, we propose Dense Cosine Similarity Maps (DCSMs) as a principled and interpretable scoring method for CLIP-like models, which solves the fundamental limitations of CLIP by retaining the semantic topology of the image patches and text tokens. This method improves upon the performance of classical CLIP-like joint encoder models on a wide array of benchmarks. We share our code and data here for reproducibility: https://github.com/Raphoo/DCSM_Ideal_CLIP

Manuel Knott, Ignacio Serna, Ethan Mann et al.

Measuring the accuracy of face recognition (FR) systems is essential for improving performance and ensuring responsible use. Accuracy is typically estimated using large annotated datasets, which are costly and difficult to obtain. We propose a novel method for 1:1 face verification that benchmarks FR systems quickly and without manual annotation, starting from approximate labels (e.g., from web search results). Unlike previous methods for training set label cleaning, ours leverages the embedding representation of the models being evaluated, achieving high accuracy in smaller-sized test datasets. Our approach reliably estimates FR accuracy and ranking, significantly reducing the time and cost of manual labeling. We also introduce the first public benchmark of five FR cloud services, revealing demographic biases, particularly lower accuracy for Asian women. Our rapid test method can democratize FR testing, promoting scrutiny and responsible use of the technology. Our method is provided as a publicly accessible tool at https://github.com/caltechvisionlab/frt-rapid-test

Rogerio Guimaraes, Frank Xiao, Pietro Perona et al.

Humans can recognize the same actions despite large context and viewpoint variations, such as differences between species (walking in spiders vs. horses), viewpoints (egocentric vs. third-person), and contexts (real life vs movies). Current deep learning models struggle with such generalization. We propose using features generated by a Vision Diffusion Model (VDM), aggregated via a transformer, to achieve human-like action recognition across these challenging conditions. We find that generalization is enhanced by the use of a model conditioned on earlier timesteps of the diffusion process to highlight semantic information over pixel level details in the extracted features. We experimentally explore the generalization properties of our approach in classifying actions across animal species, across different viewing angles, and different recording contexts. Our model sets a new state-of-the-art across all three generalization benchmarks, bringing machine action recognition closer to human-like robustness. Project page: https://www.vision.caltech.edu/actiondiff. Code: https://github.com/frankyaoxiao/ActionDiff

Héctor Carrión, Yutong Bai, Víctor A. Hernández Castro et al.

World models aim to simulate environments and enable effective agent behavior. However, modeling real-world environments presents unique challenges as they dynamically change across both space and, crucially, time. To capture these composed dynamics, we introduce a Spatio-Temporal Road Image Dataset for Exploration (STRIDE) permuting 360-degree panoramic imagery into rich interconnected observation, state and action nodes. Leveraging this structure, we can simultaneously model the relationship between egocentric views, positional coordinates, and movement commands across both space and time. We benchmark this dataset via TARDIS, a transformer-based generative world model that integrates spatial and temporal dynamics through a unified autoregressive framework trained on STRIDE. We demonstrate robust performance across a range of agentic tasks such as controllable photorealistic image synthesis, instruction following, autonomous self-control, and state-of-the-art georeferencing. These results suggest a promising direction towards sophisticated generalist agents--capable of understanding and manipulating the spatial and temporal aspects of their material environments--with enhanced embodied reasoning capabilities. Training code, datasets, and model checkpoints are made available at https://huggingface.co/datasets/Tera-AI/STRIDE.

Suzanne Stathatos, Michael Hobley, Pietro Perona et al.

Low signal-to-noise ratio videos -- such as those from underwater sonar, ultrasound, and microscopy -- pose significant challenges for computer vision models, particularly when paired clean imagery is unavailable. We present Spatiotemporal Augmentations and denoising in Video for Downstream Tasks (SAVeD), a novel self-supervised method that denoises low-SNR sensor videos using only raw noisy data. By leveraging distinctions between foreground and background motion and exaggerating objects with stronger motion signal, SAVeD enhances foreground object visibility and reduces background and camera noise without requiring clean video. SAVeD has a set of architectural optimizations that lead to faster throughput, training, and inference than existing deep learning methods. We also introduce a new denoising metric, FBD, which indicates foreground-background divergence for detection datasets without requiring clean imagery. Our approach achieves state-of-the-art results for classification, detection, tracking, and counting tasks, and it does so with fewer training resource requirements than existing deep-learning-based denoising methods. Project page: https://suzanne-stathatos.github.io/SAVeD Code page: https://github.com/suzanne-stathatos/SAVeD

Chandan Singh, Guha Balakrishnan, Pietro Perona

Measuring biases of vision systems with respect to protected attributes like gender and age is critical as these systems gain widespread use in society. However, significant correlations between attributes in benchmark datasets make it difficult to separate algorithmic bias from dataset bias. To mitigate such attribute confounding during bias analysis, we propose a matching approach that selects a subset of images from the full dataset with balanced attribute distributions across protected attributes. Our matching approach first projects real images onto a generative adversarial network (GAN)'s latent space in a manner that preserves semantic attributes. It then finds image matches in this latent space across a chosen protected attribute, yielding a dataset where semantic and perceptual attributes are balanced across the protected attribute. We validate projection and matching strategies with qualitative, quantitative, and human annotation experiments. We demonstrate our work in the context of gender bias in multiple open-source facial-recognition classifiers and find that bias persists after removing key confounders via matching. Code and documentation to reproduce the results here and apply the methods to new data is available at https://github.com/csinva/matching-with-gans .

Biagio Brattoli, Joseph Tighe, Fedor Zhdanov et al.

Trained on large datasets, deep learning (DL) can accurately classify videos into hundreds of diverse classes. However, video data is expensive to annotate. Zero-shot learning (ZSL) proposes one solution to this problem. ZSL trains a model once, and generalizes to new tasks whose classes are not present in the training dataset. We propose the first end-to-end algorithm for ZSL in video classification. Our training procedure builds on insights from recent video classification literature and uses a trainable 3D CNN to learn the visual features. This is in contrast to previous video ZSL methods, which use pretrained feature extractors. We also extend the current benchmarking paradigm: Previous techniques aim to make the test task unknown at training time but fall short of this goal. We encourage domain shift across training and test data and disallow tailoring a ZSL model to a specific test dataset. We outperform the state-of-the-art by a wide margin. Our code, evaluation procedure and model weights are available at github.com/bbrattoli/ZeroShotVideoClassification.

Neehar Kondapaneni, Oisin Mac Aodha, Pietro Perona

How do two deep neural networks differ in how they arrive at a decision? Measuring the similarity of deep networks has been a long-standing open question. Most existing methods provide a single number to measure the similarity of two networks at a given layer, but give no insight into what makes them similar or dissimilar. We introduce an interpretable representational similarity method (RSVC) to compare two networks. We use RSVC to discover shared and unique visual concepts between two models. We show that some aspects of model differences can be attributed to unique concepts discovered by one model that are not well represented in the other. Finally, we conduct extensive evaluation across different vision model architectures and training protocols to demonstrate its effectiveness.

Ira Globus-Harris, Declan Harrison, Michael Kearns et al.

Crowdsourced machine learning on competition platforms such as Kaggle is a popular and often effective method for generating accurate models. Typically, teams vie for the most accurate model, as measured by overall error on a holdout set, and it is common towards the end of such competitions for teams at the top of the leaderboard to ensemble or average their models outside the platform mechanism to get the final, best global model. In arXiv:2201.10408, the authors developed an alternative crowdsourcing framework in the context of fair machine learning, in order to integrate community feedback into models when subgroup unfairness is present and identifiable. There, unlike in classical crowdsourced ML, participants deliberately specialize their efforts by working on subproblems, such as demographic subgroups in the service of fairness. Here, we take a broader perspective on this work: we note that within this framework, participants may both specialize in the service of fairness and simply to cater to their particular expertise (e.g., focusing on identifying bird species in an image classification task). Unlike traditional crowdsourcing, this allows for the diversification of participants' efforts and may provide a participation mechanism to a larger range of individuals (e.g. a machine learning novice who has insight into a specific fairness concern). We present the first medium-scale experimental evaluation of this framework, with 46 participating teams attempting to generate models to predict income from American Community Survey data. We provide an empirical analysis of teams' approaches, and discuss the novel system architecture we developed. From here, we give concrete guidance for how best to deploy such a framework.

Ritambhara Singh, Abhishek Jain, Pietro Perona et al.

High-resolution semantic segmentation requires substantial computational resources. Traditional approaches in the field typically downscale the input images before processing and then upscale the low-resolution outputs back to their original dimensions. While this strategy effectively identifies broad regions, it often misses finer details. In this study, we demonstrate that a streamlined model capable of directly producing high-resolution segmentations can match the performance of more complex systems that generate lower-resolution results. By simplifying the network architecture, we enable the processing of images at their native resolution. Our approach leverages a bottom-up information propagation technique across various scales, which we have empirically shown to enhance segmentation accuracy. We have rigorously tested our method using leading-edge semantic segmentation datasets. Specifically, for the Cityscapes dataset, we further boost accuracy by applying the Noisy Student Training technique.

Nil-Jana Akpinar, Chia-Jung Lee, Vanessa Murdock et al.

Large Language Models (LLMs) should answer factual questions truthfully, grounded in objective knowledge, regardless of user context such as self-disclosed personal information, or system personalization. In this paper, we present the first systematic evaluation of LLM robustness to inquiry personas, i.e. user profiles that convey attributes like identity, expertise, or belief. While prior work has primarily focused on adversarial inputs or distractors for robustness testing, we evaluate plausible, human-centered inquiry persona cues that users disclose in real-world interactions. We find that such cues can meaningfully alter QA accuracy and trigger failure modes such as refusals, hallucinated limitations, and role confusion. These effects highlight how model sensitivity to user framing can compromise factual reliability, and position inquiry persona testing as an effective tool for robustness evaluation.

Yue Song, T. Anderson Keller, Sevan Brodjian et al.

Orientation-rich images, such as fingerprints and textures, often exhibit coherent angular directional patterns that are challenging to model using standard generative approaches based on isotropic Euclidean diffusion. Motivated by the role of phase synchronization in biological systems, we propose a score-based generative model built on periodic domains by leveraging stochastic Kuramoto dynamics in the diffusion process. In neural and physical systems, Kuramoto models capture synchronization phenomena across coupled oscillators -- a behavior that we re-purpose here as an inductive bias for structured image generation. In our framework, the forward process performs \textit{synchronization} among phase variables through globally or locally coupled oscillator interactions and attraction to a global reference phase, gradually collapsing the data into a low-entropy von Mises distribution. The reverse process then performs \textit{desynchronization}, generating diverse patterns by reversing the dynamics with a learned score function. This approach enables structured destruction during forward diffusion and a hierarchical generation process that progressively refines global coherence into fine-scale details. We implement wrapped Gaussian transition kernels and periodicity-aware networks to account for the circular geometry. Our method achieves competitive results on general image benchmarks and significantly improves generation quality on orientation-dense datasets like fingerprints and textures. Ultimately, this work demonstrates the promise of biologically inspired synchronization dynamics as structured priors in generative modeling.

Yue Song, T. Anderson Keller, Yisong Yue et al.

Neural populations exhibit latent dynamical structures that drive time-evolving spiking activities, motivating the search for models that capture both intrinsic network dynamics and external unobserved influences. In this work, we introduce LangevinFlow, a sequential Variational Auto-Encoder where the time evolution of latent variables is governed by the underdamped Langevin equation. Our approach incorporates physical priors -- such as inertia, damping, a learned potential function, and stochastic forces -- to represent both autonomous and non-autonomous processes in neural systems. Crucially, the potential function is parameterized as a network of locally coupled oscillators, biasing the model toward oscillatory and flow-like behaviors observed in biological neural populations. Our model features a recurrent encoder, a one-layer Transformer decoder, and Langevin dynamics in the latent space. Empirically, our method outperforms state-of-the-art baselines on synthetic neural populations generated by a Lorenz attractor, closely matching ground-truth firing rates. On the Neural Latents Benchmark (NLB), the model achieves superior held-out neuron likelihoods (bits per spike) and forward prediction accuracy across four challenging datasets. It also matches or surpasses alternative methods in decoding behavioral metrics such as hand velocity. Overall, this work introduces a flexible, physics-inspired, high-performing framework for modeling complex neural population dynamics and their unobserved influences.

Neehar Kondapaneni, Oisin Mac Aodha, Pietro Perona

We propose a method for discovering and visualizing the differences between two learned representations, enabling more direct and interpretable model comparisons. We validate our method, which we call Representational Differences Explanations (RDX), by using it to compare models with known conceptual differences and demonstrate that it recovers meaningful distinctions where existing explainable AI (XAI) techniques fail. Applied to state-of-the-art models on challenging subsets of the ImageNet and iNaturalist datasets, RDX reveals both insightful representational differences and subtle patterns in the data. Although comparison is a cornerstone of scientific analysis, current tools in machine learning, namely post hoc XAI methods, struggle to support model comparison effectively. Our work addresses this gap by introducing an effective and explainable tool for contrasting model representations.

Kai Van Brunt, Justin Kay, Timm Haucke et al. · mit

Accurate estimates of salmon escapement - the number of fish migrating upstream to spawn - are key data for conservation and fishery management. Existing methods for salmon counting using high-resolution imaging sonar hardware are non-invasive and compatible with computer vision processing. Prior work in this area has utilized object detection and tracking based methods for automated salmon counting. However, these techniques remain inaccessible to many sonar deployment sites due to limited compute and connectivity in the field. We propose an alternative lightweight computer vision method for fish counting based on analyzing echograms - temporal representations that compress several hundred frames of imaging sonar video into a single image. We predict upstream and downstream counts within 200-frame time windows directly from echograms using a ResNet-18 model, and propose a set of domain-specific image augmentations and a weakly-supervised training protocol to further improve results. We achieve a count error of 23% on representative data from the Kenai River in Alaska, demonstrating the feasibility of our approach.

Riccardo Fogliato, Pratik Patil, Mathew Monfort et al.

Model performance evaluation is a critical and expensive task in machine learning and computer vision. Without clear guidelines, practitioners often estimate model accuracy using a one-time completely random selection of the data. However, by employing tailored sampling and estimation strategies, one can obtain more precise estimates and reduce annotation costs. In this paper, we propose a statistical framework for model evaluation that includes stratification, sampling, and estimation components. We examine the statistical properties of each component and evaluate their efficiency (precision). One key result of our work is that stratification via k-means clustering based on accurate predictions of model performance yields efficient estimators. Our experiments on computer vision datasets show that this method consistently provides more precise accuracy estimates than the traditional simple random sampling, even with substantial efficiency gains of 10x. We also find that model-assisted estimators, which leverage predictions of model accuracy on the unlabeled portion of the dataset, are generally more efficient than the traditional estimates based solely on the labeled data.

Neehar Kondapaneni, Markus Marks, Oisin Mac Aodha et al.

We introduce Discovering Conceptual Network Explanations (DCNE), a new approach for generating human-comprehensible visual explanations to enhance the interpretability of deep neural image classifiers. Our method automatically finds visual explanations that are critical for discriminating between classes. This is achieved by simultaneously optimizing three criteria: the explanations should be few, diverse, and human-interpretable. Our approach builds on the recently introduced Concept Relevance Propagation (CRP) explainability method. While CRP is effective at describing individual neuronal activations, it generates too many concepts, which impacts human comprehension. Instead, DCNE selects the few most important explanations. We introduce a new evaluation dataset centered on the challenging task of classifying birds, enabling us to compare the alignment of DCNE's explanations to those of human expert-defined ones. Compared to existing eXplainable Artificial Intelligence (XAI) methods, DCNE has a desirable trade-off between conciseness and completeness when summarizing network explanations. It produces 1/30 of CRP's explanations while only resulting in a slight reduction in explanation quality. DCNE represents a step forward in making neural network decisions accessible and interpretable to humans, providing a valuable tool for both researchers and practitioners in XAI and model alignment.

Julio Arroyo, Pietro Perona, Elijah Cole

Annotating data for multi-label classification is prohibitively expensive because every category of interest must be confirmed to be present or absent. Recent work on single positive multi-label (SPML) learning shows that it is possible to train effective multi-label classifiers using only one positive label per image. However, the standard benchmarks for SPML are derived from traditional multi-label classification datasets by retaining one positive label for each training example (chosen uniformly at random) and discarding all other labels. In realistic settings it is not likely that positive labels are chosen uniformly at random. This work introduces protocols for studying label bias in SPML and provides new empirical results.

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.

Yair Kittenplon, Inbal Lavi, Sharon Fogel et al.

Text spotting end-to-end methods have recently gained attention in the literature due to the benefits of jointly optimizing the text detection and recognition components. Existing methods usually have a distinct separation between the detection and recognition branches, requiring exact annotations for the two tasks. We introduce TextTranSpotter (TTS), a transformer-based approach for text spotting and the first text spotting framework which may be trained with both fully- and weakly-supervised settings. By learning a single latent representation per word detection, and using a novel loss function based on the Hungarian loss, our method alleviates the need for expensive localization annotations. Trained with only text transcription annotations on real data, our weakly-supervised method achieves competitive performance with previous state-of-the-art fully-supervised methods. When trained in a fully-supervised manner, TextTranSpotter shows state-of-the-art results on multiple benchmarks.

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.

Jennifer J. Sun, Serim Ryou, Roni Goldshmid et al.

We propose a method for learning the posture and structure of agents from unlabelled behavioral videos. Starting from the observation that behaving agents are generally the main sources of movement in behavioral videos, our method, Behavioral Keypoint Discovery (B-KinD), uses an encoder-decoder architecture with a geometric bottleneck to reconstruct the spatiotemporal difference between video frames. By focusing only on regions of movement, our approach works directly on input videos without requiring manual annotations. Experiments on a variety of agent types (mouse, fly, human, jellyfish, and trees) demonstrate the generality of our approach and reveal that our discovered keypoints represent semantically meaningful body parts, which achieve state-of-the-art performance on keypoint regression among self-supervised methods. Additionally, B-KinD achieve comparable performance to supervised keypoints on downstream tasks, such as behavior classification, suggesting that our method can dramatically reduce model training costs vis-a-vis supervised methods.

Serim Ryou, Pietro Perona

In this paper, we propose a method for keypoint discovery from a 2D image using image-level supervision. Recent works on unsupervised keypoint discovery reliably discover keypoints of aligned instances. However, when the target instances have high viewpoint or appearance variation, the discovered keypoints do not match the semantic correspondences over different images. Our work aims to discover keypoints even when the target instances have high viewpoint and appearance variation by using image-level supervision. Motivated by the weakly-supervised learning approach, our method exploits image-level supervision to identify discriminative parts and infer the viewpoint of the target instance. To discover diverse parts, we adopt a conditional image generation approach using a pair of images with structural deformation. Finally, we enforce a viewpoint-based equivariance constraint using the keypoints from the image-level supervision to resolve the spatial correlation problem that consistently appears in the images taken from various viewpoints. Our approach achieves state-of-the-art performance for the task of keypoint estimation on the limited supervision scenarios. Furthermore, the discovered keypoints are directly applicable to downstream tasks without requiring any keypoint labels.

Sara Beery, Elijah Cole, Joseph Parker et al.

Conservation science depends on an accurate understanding of what's happening in a given ecosystem. How many species live there? What is the makeup of the population? How is that changing over time? Species Distribution Modeling (SDM) seeks to predict the spatial (and sometimes temporal) patterns of species occurrence, i.e. where a species is likely to be found. The last few years have seen a surge of interest in applying powerful machine learning tools to challenging problems in ecology. Despite its considerable importance, SDM has received relatively little attention from the computer science community. Our goal in this work is to provide computer scientists with the necessary background to read the SDM literature and develop ecologically useful ML-based SDM algorithms. In particular, we introduce key SDM concepts and terminology, review standard models, discuss data availability, and highlight technical challenges and pitfalls.

Elijah Cole, Oisin Mac Aodha, Titouan Lorieul et al.

Predicting all applicable labels for a given image is known as multi-label classification. Compared to the standard multi-class case (where each image has only one label), it is considerably more challenging to annotate training data for multi-label classification. When the number of potential labels is large, human annotators find it difficult to mention all applicable labels for each training image. Furthermore, in some settings detection is intrinsically difficult e.g. finding small object instances in high resolution images. As a result, multi-label training data is often plagued by false negatives. We consider the hardest version of this problem, where annotators provide only one relevant label for each image. As a result, training sets will have only one positive label per image and no confirmed negatives. We explore this special case of learning from missing labels across four different multi-label image classification datasets for both linear classifiers and end-to-end fine-tuned deep networks. We extend existing multi-label losses to this setting and propose novel variants that constrain the number of expected positive labels during training. Surprisingly, we show that in some cases it is possible to approach the performance of fully labeled classifiers despite training with significantly fewer confirmed labels.

Jennifer J. Sun, Tomomi Karigo, Dipam Chakraborty et al.

Multi-agent behavior modeling aims to understand the interactions that occur between agents. We present a multi-agent dataset from behavioral neuroscience, the Caltech Mouse Social Interactions (CalMS21) Dataset. Our dataset consists of trajectory data of social interactions, recorded from videos of freely behaving mice in a standard resident-intruder assay. To help accelerate behavioral studies, the CalMS21 dataset provides benchmarks to evaluate the performance of automated behavior classification methods in three settings: (1) for training on large behavioral datasets all annotated by a single annotator, (2) for style transfer to learn inter-annotator differences in behavior definitions, and (3) for learning of new behaviors of interest given limited training data. The dataset consists of 6 million frames of unlabeled tracked poses of interacting mice, as well as over 1 million frames with tracked poses and corresponding frame-level behavior annotations. The challenge of our dataset is to be able to classify behaviors accurately using both labeled and unlabeled tracking data, as well as being able to generalize to new settings.

Aditya Deshpande, Alessandro Achille, Avinash Ravichandran et al.

Fine-tuning from a collection of models pre-trained on different domains (a "model zoo") is emerging as a technique to improve test accuracy in the low-data regime. However, model selection, i.e. how to pre-select the right model to fine-tune from a model zoo without performing any training, remains an open topic. We use a linearized framework to approximate fine-tuning, and introduce two new baselines for model selection -- Label-Gradient and Label-Feature Correlation. Since all model selection algorithms in the literature have been tested on different use-cases and never compared directly, we introduce a new comprehensive benchmark for model selection comprising of: i) A model zoo of single and multi-domain models, and ii) Many target tasks. Our benchmark highlights accuracy gain with model zoo compared to fine-tuning Imagenet models. We show our model selection baseline can select optimal models to fine-tune in few selections and has the highest ranking correlation to fine-tuning accuracy compared to existing algorithms.

Aviad Aberdam, Ron Litman, Shahar Tsiper et al.

We propose a framework for sequence-to-sequence contrastive learning (SeqCLR) of visual representations, which we apply to text recognition. To account for the sequence-to-sequence structure, each feature map is divided into different instances over which the contrastive loss is computed. This operation enables us to contrast in a sub-word level, where from each image we extract several positive pairs and multiple negative examples. To yield effective visual representations for text recognition, we further suggest novel augmentation heuristics, different encoder architectures and custom projection heads. Experiments on handwritten text and on scene text show that when a text decoder is trained on the learned representations, our method outperforms non-sequential contrastive methods. In addition, when the amount of supervision is reduced, SeqCLR significantly improves performance compared with supervised training, and when fine-tuned with 100% of the labels, our method achieves state-of-the-art results on standard handwritten text recognition benchmarks.

Neehar Kondapaneni, Pietro Perona

The ability to understand and manipulate numbers and quantities emerges during childhood, but the mechanism through which humans acquire and develop this ability is still poorly understood. We explore this question through a model, assuming that the learner is able to pick up and place small objects from, and to, locations of its choosing, and will spontaneously engage in such undirected manipulation. We further assume that the learner's visual system will monitor the changing arrangements of objects in the scene and will learn to predict the effects of each action by comparing perception with a supervisory signal from the motor system. We model perception using standard deep networks for feature extraction and classification, and gradient descent learning. Our main finding is that, from learning the task of action prediction, an unexpected image representation emerges exhibiting regularities that foreshadow the perception and representation of numbers and quantity. These include distinct categories for zero and the first few natural numbers, a strict ordering of the numbers, and a one-dimensional signal that correlates with numerical quantity. As a result, our model acquires the ability to estimate numerosity, i.e. the number of objects in the scene, as well as subitization, i.e. the ability to recognize at a glance the exact number of objects in small scenes. Remarkably, subitization and numerosity estimation extrapolate to scenes containing many objects, far beyond the three objects used during training. We conclude that important aspects of a facility with numbers and quantities may be learned with supervision from a simple pre-training task. Our observations suggest that cross-modal learning is a powerful learning mechanism that may be harnessed in artificial intelligence.

Jennifer J. Sun, Ann Kennedy, Eric Zhan et al.

Specialized domain knowledge is often necessary to accurately annotate training sets for in-depth analysis, but can be burdensome and time-consuming to acquire from domain experts. This issue arises prominently in automated behavior analysis, in which agent movements or actions of interest are detected from video tracking data. To reduce annotation effort, we present TREBA: a method to learn annotation-sample efficient trajectory embedding for behavior analysis, based on multi-task self-supervised learning. The tasks in our method can be efficiently engineered by domain experts through a process we call "task programming", which uses programs to explicitly encode structured knowledge from domain experts. Total domain expert effort can be reduced by exchanging data annotation time for the construction of a small number of programmed tasks. We evaluate this trade-off using data from behavioral neuroscience, in which specialized domain knowledge is used to identify behaviors. We present experimental results in three datasets across two domains: mice and fruit flies. Using embeddings from TREBA, we reduce annotation burden by up to a factor of 10 without compromising accuracy compared to state-of-the-art features. Our results thus suggest that task programming and self-supervision can be an effective way to reduce annotation effort for domain experts.

Guha Balakrishnan, Yuanjun Xiong, Wei Xia et al.

Measuring algorithmic bias is crucial both to assess algorithmic fairness, and to guide the improvement of algorithms. Current methods to measure algorithmic bias in computer vision, which are based on observational datasets, are inadequate for this task because they conflate algorithmic bias with dataset bias. To address this problem we develop an experimental method for measuring algorithmic bias of face analysis algorithms, which manipulates directly the attributes of interest, e.g., gender and skin tone, in order to reveal causal links between attribute variation and performance change. Our proposed method is based on generating synthetic ``transects'' of matched sample images that are designed to differ along specific attributes while leaving other attributes constant. A crucial aspect of our approach is relying on the perception of human observers, both to guide manipulations, and to measure algorithmic bias. Besides allowing the measurement of algorithmic bias, synthetic transects have other advantages with respect to observational datasets: they sample attributes more evenly allowing for more straightforward bias analysis on minority and intersectional groups, they enable prediction of bias in new scenarios, they greatly reduce ethical and legal challenges, and they are economical and fast to obtain, helping make bias testing affordable and widely available. We validate our method by comparing it to a study that employs the traditional observational method for analyzing bias in gender classification algorithms. The two methods reach different conclusions. While the observational method reports gender and skin color biases, the experimental method reveals biases due to gender, hair length, age, and facial hair.

Daniel Laumer, Nico Lang, Natalie van Doorn et al.

We introduce an approach for updating older tree inventories with geographic coordinates using street-level panorama images and a global optimization framework for tree instance matching. Geolocations of trees in inventories until the early 2000s where recorded using street addresses whereas newer inventories use GPS. Our method retrofits older inventories with geographic coordinates to allow connecting them with newer inventories to facilitate long-term studies on tree mortality etc. What makes this problem challenging is the different number of trees per street address, the heterogeneous appearance of different tree instances in the images, ambiguous tree positions if viewed from multiple images and occlusions. To solve this assignment problem, we (i) detect trees in Google street-view panoramas using deep learning, (ii) combine multi-view detections per tree into a single representation, (iii) and match detected trees with given trees per street address with a global optimization approach. Experiments for > 50000 trees in 5 cities in California, USA, show that we are able to assign geographic coordinates to 38 % of the street trees, which is a good starting point for long-term studies on the ecosystem services value of street trees at large scale.

Bhairav Chidambaram, Mason McGill, Pietro Perona

We design an agent to search for frames of interest in video stored on a remote server, under bandwidth constraints. Using a convolutional neural network to score individual frames and a hidden Markov model to propagate predictions across frames, our agent accurately identifies temporal regions of interest based on sparse, strategically sampled frames. On a subset of the ImageNet-VID dataset, we demonstrate that using a hidden Markov model to interpolate between frame scores allows requests of 98% of frames to be omitted, without compromising frame-of-interest classification accuracy.

Yang Liu, Pietro Perona, Markus Meister

The recently proposed panoptic segmentation task presents a significant challenge of image understanding with computer vision by unifying semantic segmentation and instance segmentation tasks. In this paper we present an efficient and novel panoptic data augmentation (PanDA) method which operates exclusively in pixel space, requires no additional data or training, and is computationally cheap to implement. By retraining original state-of-the-art models on PanDA augmented datasets generated with a single frozen set of parameters, we show robust performance gains in panoptic segmentation, instance segmentation, as well as detection across models, backbones, dataset domains, and scales. Finally, the effectiveness of unrealistic-looking training images synthesized by PanDA suggest that one should rethink the need for image realism for efficient data augmentation.

Steve Branson, Jan Dirk Wegner, David Hall et al.

Up-to-date catalogs of the urban tree population are important for municipalities to monitor and improve quality of life in cities. Despite much research on automation of tree mapping, mainly relying on dedicated airborne LiDAR or hyperspectral campaigns, trees are still mostly mapped manually in practice. We present a fully automated tree detection and species recognition pipeline to process thousands of trees within a few hours using publicly available aerial and street view images of Google MapsTM. These data provide rich information (viewpoints, scales) from global tree shapes to bark textures. Our work-flow is built around a supervised classification that automatically learns the most discriminative features from thousands of trees and corresponding, public tree inventory data. In addition, we introduce a change tracker to keep urban tree inventories up-to-date. Changes of individual trees are recognized at city-scale by comparing street-level images of the same tree location at two different times. Drawing on recent advances in computer vision and machine learning, we apply convolutional neural networks (CNN) for all classification tasks. We propose the following pipeline: download all available panoramas and overhead images of an area of interest, detect trees per image and combine multi-view detections in a probabilistic framework, adding prior knowledge; recognize fine-grained species of detected trees. In a later, separate module, track trees over time and identify the type of change. We believe this is the first work to exploit publicly available image data for fine-grained tree mapping at city-scale, respectively over many thousands of trees. Experiments in the city of Pasadena, California, USA show that we can detect > 70% of the street trees, assign correct species to > 80% for 40 different species, and correctly detect and classify changes in > 90% of the cases.

Serim Ryou, Seong-Gyun Jeong, Pietro Perona

We propose a novel loss function that dynamically rescales the cross entropy based on prediction difficulty regarding a sample. Deep neural network architectures in image classification tasks struggle to disambiguate visually similar objects. Likewise, in human pose estimation symmetric body parts often confuse the network with assigning indiscriminative scores to them. This is due to the output prediction, in which only the highest confidence label is selected without taking into consideration a measure of uncertainty. In this work, we define the prediction difficulty as a relative property coming from the confidence score gap between positive and negative labels. More precisely, the proposed loss function penalizes the network to avoid the score of a false prediction being significant. To demonstrate the efficacy of our loss function, we evaluate it on two different domains: image classification and human pose estimation. We find improvements in both applications by achieving higher accuracy compared to the baseline methods.