Subhransu Maji

Gopal Sharma, Kangxue Yin, Subhransu Maji et al.

We propose to utilize self-supervised techniques in the 2D domain for fine-grained 3D shape segmentation tasks. This is inspired by the observation that view-based surface representations are more effective at modeling high-resolution surface details and texture than their 3D counterparts based on point clouds or voxel occupancy. Specifically, given a 3D shape, we render it from multiple views, and set up a dense correspondence learning task within the contrastive learning framework. As a result, the learned 2D representations are view-invariant and geometrically consistent, leading to better generalization when trained on a limited number of labeled shapes compared to alternatives that utilize self-supervision in 2D or 3D alone. Experiments on textured (RenderPeople) and untextured (PartNet) 3D datasets show that our method outperforms state-of-the-art alternatives in fine-grained part segmentation. The improvements over baselines are greater when only a sparse set of views is available for training or when shapes are textured, indicating that MvDeCor benefits from both 2D processing and 3D geometric reasoning.

Zezhou Cheng, Carlos Esteves, Varun Jampani et al.

A critical obstacle preventing NeRF models from being deployed broadly in the wild is their reliance on accurate camera poses. Consequently, there is growing interest in extending NeRF models to jointly optimize camera poses and scene representation, which offers an alternative to off-the-shelf SfM pipelines which have well-understood failure modes. Existing approaches for unposed NeRF operate under limited assumptions, such as a prior pose distribution or coarse pose initialization, making them less effective in a general setting. In this work, we propose a novel approach, LU-NeRF, that jointly estimates camera poses and neural radiance fields with relaxed assumptions on pose configuration. Our approach operates in a local-to-global manner, where we first optimize over local subsets of the data, dubbed mini-scenes. LU-NeRF estimates local pose and geometry for this challenging few-shot task. The mini-scene poses are brought into a global reference frame through a robust pose synchronization step, where a final global optimization of pose and scene can be performed. We show our LU-NeRF pipeline outperforms prior attempts at unposed NeRF without making restrictive assumptions on the pose prior. This allows us to operate in the general SE(3) pose setting, unlike the baselines. Our results also indicate our model can be complementary to feature-based SfM pipelines as it compares favorably to COLMAP on low-texture and low-resolution images.

Zezhou Cheng, Menglei Chai, Jian Ren et al.

Creating and editing the shape and color of 3D objects require tremendous human effort and expertise. Compared to direct manipulation in 3D interfaces, 2D interactions such as sketches and scribbles are usually much more natural and intuitive for the users. In this paper, we propose a generic multi-modal generative model that couples the 2D modalities and implicit 3D representations through shared latent spaces. With the proposed model, versatile 3D generation and manipulation are enabled by simply propagating the editing from a specific 2D controlling modality through the latent spaces. For example, editing the 3D shape by drawing a sketch, re-colorizing the 3D surface via painting color scribbles on the 2D rendering, or generating 3D shapes of a certain category given one or a few reference images. Unlike prior works, our model does not require re-training or fine-tuning per editing task and is also conceptually simple, easy to implement, robust to input domain shifts, and flexible to diverse reconstruction on partial 2D inputs. We evaluate our framework on two representative 2D modalities of grayscale line sketches and rendered color images, and demonstrate that our method enables various shape manipulation and generation tasks with these 2D modalities.

Oindrila Saha, Zezhou Cheng, Subhransu Maji

A significant bottleneck in training deep networks for part segmentation is the cost of obtaining detailed annotations. We propose a framework to exploit coarse labels such as figure-ground masks and keypoint locations that are readily available for some categories to improve part segmentation models. A key challenge is that these annotations were collected for different tasks and with different labeling styles and cannot be readily mapped to the part labels. To this end, we propose to jointly learn the dependencies between labeling styles and the part segmentation model, allowing us to utilize supervision from diverse labels. To evaluate our approach we develop a benchmark on the Caltech-UCSD birds and OID Aircraft dataset. Our approach outperforms baselines based on multi-task learning, semi-supervised learning, and competitive methods relying on loss functions manually designed to exploit sparse-supervision.

Chenyun Wu, Subhransu Maji

We investigate how well CLIP understands texture in natural images described by natural language. To this end, we analyze CLIP's ability to: (1) perform zero-shot learning on various texture and material classification datasets; (2) represent compositional properties of texture such as red dots or yellow stripes on the Describable Texture in Detail(DTDD) dataset; and (3) aid fine-grained categorization of birds in photographs described by color and texture of their body parts.

Gustavo Perez, Subhransu Maji, Daniel Sheldon

Many modern applications use computer vision to detect and count objects in massive image collections. However, when the detection task is very difficult or in the presence of domain shifts, the counts may be inaccurate even with significant investments in training data and model development. We propose DISCount -- a detector-based importance sampling framework for counting in large image collections that integrates an imperfect detector with human-in-the-loop screening to produce unbiased estimates of counts. We propose techniques for solving counting problems over multiple spatial or temporal regions using a small number of screened samples and estimate confidence intervals. This enables end-users to stop screening when estimates are sufficiently accurate, which is often the goal in a scientific study. On the technical side we develop variance reduction techniques based on control variates and prove the (conditional) unbiasedness of the estimators. DISCount leads to a 9-12x reduction in the labeling costs over naive screening for tasks we consider, such as counting birds in radar imagery or estimating damaged buildings in satellite imagery, and also surpasses alternative covariate-based screening approaches in efficiency.

Oindrila Saha, Subhransu Maji

We develop techniques for refining representations for fine-grained classification and segmentation tasks in a self-supervised manner. We find that fine-tuning methods based on instance-discriminative contrastive learning are not as effective, and posit that recognizing part-specific variations is crucial for fine-grained categorization. We present an iterative learning approach that incorporates part-centric equivariance and invariance objectives. First, pixel representations are clustered to discover parts. We analyze the representations from convolutional and vision transformer networks that are best suited for this task. Then, a part-centric learning step aggregates and contrasts representations of parts within an image. We show that this improves the performance on image classification and part segmentation tasks across datasets. For example, under a linear-evaluation scheme, the classification accuracy of a ResNet50 trained on ImageNet using DetCon, a self-supervised learning approach, improves from 35.4% to 42.0% on the Caltech-UCSD Birds, from 35.5% to 44.1% on the FGVC Aircraft, and from 29.7% to 37.4% on the Stanford Cars. We also observe significant gains in few-shot part segmentation tasks using the proposed technique, while instance-discriminative learning was not as effective. Smaller, yet consistent, improvements are also observed for stronger networks based on transformers.

Zezhou Cheng, Matheus Gadelha, Subhransu Maji

We propose a technique for learning single-view 3D object pose estimation models by utilizing a new source of data -- in-the-wild videos where objects turn. Such videos are prevalent in practice (e.g., cars in roundabouts, airplanes near runways) and easy to collect. We show that classical structure-from-motion algorithms, coupled with the recent advances in instance detection and feature matching, provides surprisingly accurate relative 3D pose estimation on such videos. We propose a multi-stage training scheme that first learns a canonical pose across a collection of videos and then supervises a model for single-view pose estimation. The proposed technique achieves competitive performance with respect to existing state-of-the-art on standard benchmarks for 3D pose estimation, without requiring any pose labels during training. We also contribute an Accidental Turntables Dataset, containing a challenging set of 41,212 images of cars in cluttered backgrounds, motion blur and illumination changes that serves as a benchmark for 3D pose estimation.

Logan Lawrence, Mustafa Chasmai, Rangel Daroya et al.

Fine-grained bird species identification in the wild is frequently unanswerable from a single image: key cues may be non-visual (e.g. vocalization), or obscured due to occlusion, camera angle, or low resolution. Yet today's multimodal systems are typically judged on answerable, in-schema cases, encouraging confident guesses rather than principled abstention. We propose the RealBirdID benchmark: given an image of a bird, a system should either answer with a species or abstain with a concrete, evidence-based rationale: "requires vocalization," "low quality image," or "view obstructed". For each genus, the dataset includes a validation split composed of curated unanswerable examples with labeled rationales, paired with a companion set of clearly answerable instances. We find that (1) the species identification on the answerable set is challenging for a variety of open-source and proprietary models (less than 13% accuracy for MLLMs including GPT-5 and Gemini-2.5 Pro), (2) models with greater classification ability are not necessarily more calibrated to abstain from unanswerable examples, and (3) that MLLMs generally fail at providing correct reasons even when they do abstain. RealBirdID establishes a focused target for abstention-aware fine-grained recognition and a recipe for measuring progress.

Rangel Daroya, Aaron Sun, Subhransu Maji

We present a set of metrics that utilize vision priors to effectively assess the performance of saliency methods on image classification tasks. To understand behavior in deep learning models, many methods provide visual saliency maps emphasizing image regions that most contribute to a model prediction. However, there is limited work on analyzing the reliability of saliency methods in explaining model decisions. We propose the metric COnsistency-SEnsitivity (COSE) that quantifies the equivariant and invariant properties of visual model explanations using simple data augmentations. Through our metrics, we show that although saliency methods are thought to be architecture-independent, most methods could better explain transformer-based models over convolutional-based models. In addition, GradCAM was found to outperform other methods in terms of COSE but was shown to have limitations such as lack of variability for fine-grained datasets. The duality between consistency and sensitivity allow the analysis of saliency methods from different angles. Ultimately, we find that it is important to balance these two metrics for a saliency map to faithfully show model behavior.

Oindrila Saha, Vojtech Krs, Radomir Mech et al.

Recent progress in large language models (LLMs) has shown that reasoning improves when intermediate thoughts are externalized into explicit workspaces, such as chain-of-thought traces or tool-augmented reasoning. Yet, visual language models (VLMs) lack an analogous mechanism for spatial reasoning, limiting their ability to generate images that accurately reflect geometric relations, object identities, and compositional intent. We introduce the concept of a spatial scratchpad -- a 3D reasoning substrate that bridges linguistic intent and image synthesis. Given a text prompt, our framework parses subjects and background elements, instantiates them as editable 3D meshes, and employs agentic scene planning for placement, orientation, and viewpoint selection. The resulting 3D arrangement is rendered back into the image domain with identity-preserving cues, enabling the VLM to generate spatially consistent and visually coherent outputs. Unlike prior 2D layout-based methods, our approach supports intuitive 3D edits that propagate reliably into final images. Empirically, it achieves a 32% improvement in text alignment on GenAI-Bench, demonstrating the benefit of explicit 3D reasoning for precise, controllable image generation. Our results highlight a new paradigm for vision-language models that deliberate not only in language, but also in space. Code and visualizations at https://oindrilasaha.github.io/3DScratchpad/

Oindrila Saha, Grant Van Horn, Subhransu Maji

The zero-shot performance of existing vision-language models (VLMs) such as CLIP is limited by the availability of large-scale, aligned image and text datasets in specific domains. In this work, we leverage two complementary sources of information -- descriptions of categories generated by large language models (LLMs) and abundant, fine-grained image classification datasets -- to improve the zero-shot classification performance of VLMs across fine-grained domains. On the technical side, we develop methods to train VLMs with this "bag-level" image-text supervision. We find that simply using these attributes at test-time does not improve performance, but our training strategy, for example, on the iNaturalist dataset, leads to an average improvement of 4-5% in zero-shot classification accuracy for novel categories of birds and flowers. Similar improvements are observed in domains where a subset of the categories was used to fine-tune the model. By prompting LLMs in various ways, we generate descriptions that capture visual appearance, habitat, and geographic regions and pair them with existing attributes such as the taxonomic structure of the categories. We systematically evaluate their ability to improve zero-shot categorization in natural domains. Our findings suggest that geographic priors can be just as effective and are complementary to visual appearance. Our method also outperforms prior work on prompt-based tuning of VLMs. We release the benchmark, consisting of 14 datasets at https://github.com/cvl-umass/AdaptCLIPZS , which will contribute to future research in zero-shot recognition.

Tung-I Chen, Lingdong Wang, Subhransu Maji et al.

Volumetric media promises next-generation content delivery applications, but its bandwidth demand remains a key bottleneck. Implicit and hybrid volumetric representations reduce model sizes, yet still require careful coding to reach 2D video-like bitrates. We present CATRF, a standard-codec-in-the-loop compression framework for plane-factorized radiance fields. During training, we quantize and pack 2D feature planes into codec-friendly canvases, run a standard codec roundtrip (JPEG/VP9/HEVC/AV1), then unpack and dequantize the decoded features before volume rendering. We use a straight-through estimator (STE) to insert the non-differentiable, standard codec pipeline into the training loop, allowing radiance-field features to adapt directly to the real, client-side codec distortions without introducing any learned codec parameters. On both static and dynamic benchmarks, CATRF consistently achieves a better rate-distortion trade-off over codec-agnostic and learned-codec-in-the-loop baselines, and also outperforms recent compressed 3DGS methods in both compression efficiency and decoding speed. These results highlight a practical path toward low-bitrate, compression-resilient volumetric representations for free-viewpoint video streaming.

Max Hamilton, Daniel Sheldon, Subhransu Maji

Two-point correlation functions (2PCF) are widely used to characterize how points cluster in space. In this work, we study the problem of measuring the 2PCF over a large set of points, restricted to a subset satisfying a property of interest. An example comes from astronomy, where scientists measure the 2PCF of star clusters, which make up only a tiny subset of possible sources within a galaxy. This task typically requires careful labeling of sources to construct catalogs, which is time-consuming. We present a human-in-the-loop framework for efficient estimation of 2PCF of target sources. By leveraging a pre-trained classifier to guide sampling, our approach adaptively selects the most informative points for human annotation. After each annotation, it produces unbiased estimates of pair counts across multiple distance bins simultaneously. Compared to simple Monte Carlo approaches, our method achieves substantially lower variance while significantly reducing annotation effort. We introduce a novel unbiased estimator, sampling strategy, and confidence interval construction that together enable scalable and statistically grounded measurement of two-point correlations in astronomy datasets.

Aaron Sun, Oindrila Saha, Subhransu Maji

Since the advent of controllable image generation, increasingly rich modes of control have enabled greater customization and accessibility for everyday users. Zero-shot, identity-preserving models such as Insert Anything and OminiControl now support applications like virtual try-on without requiring additional fine-tuning. While these models may be fitting for humans and rigid everyday objects, they still have limitations for non-rigid or fine-grained categories. These domains often lack accessible, high-quality data -- especially videos or multi-view observations of the same subject -- making them difficult both to evaluate and to improve upon. Yet, such domains are essential for moving beyond content creation toward applications that demand accuracy and fine detail. Birds are an excellent domain for this task: they exhibit high diversity, require fine-grained cues for identification, and come in a wide variety of poses. We introduce the NABirds Look-Alikes (NABLA) dataset, consisting of 4,759 expert-curated image pairs. Together with 1,073 pairs collected from multi-image observations on iNaturalist and a small set of videos, this forms a benchmark for evaluating identity-preserving generation of birds. We show that state-of-the-art baselines fail to maintain identity on this dataset, and we demonstrate that training on images grouped by species, age, and sex -- used as a proxy for identity -- substantially improves performance on both seen and unseen species.

Wuao Liu, Mustafa Chasmai, Subhransu Maji et al.

Bioacoustic recognition requires fine-grained acoustic understanding to distinguish similar-sounding species. However, many large-scale data repositories such as iNaturalist are weakly annotated, often with only a single positive species label per recording, making supervised learning particularly challenging. Inspired by advances in computer vision, recent approaches have shifted toward self-supervised learning to capture the underlying structure of audio without relying on exhaustive annotations. In particular, masked autoencoders (MAE) have shown strong transferability on massive audio corpora, yet their effectiveness in more modest bioacoustic settings remains underexplored. In this work, we conduct a systematic study of MAE pretraining for species classification on iNatSounds, analyzing the impacts of pretraining data scale, domain specificity, data curation, and transfer strategies. Consistent with prior work, we find that models pretrained on diverse general audio data achieve the best transfer performance on iNatSounds. Contrary to observations from large-scale audio benchmarks, we find that (1) additional masked reconstruction pretraining on domain-specific data provides limited benefits and may even degrade performance relative to off-the-shelf models, and (2) selective data filtering offers a negligible advantage when the overall data scale is limited. Our results indicate that, in moderate-sized fine-grained bioacoustic settings, pretraining scale dominates objective design. These findings further clarify when MAE-based pretraining is effective and provide practical guidance for model selection under limited supervision.

Justin Kay, Grant Van Horn, Subhransu Maji et al. · mit

The widespread availability of off-the-shelf machine learning models poses a challenge: which model, of the many available candidates, should be chosen for a given data analysis task? This question of model selection is traditionally answered by collecting and annotating a validation dataset -- a costly and time-intensive process. We propose a method for active model selection, using predictions from candidate models to prioritize the labeling of test data points that efficiently differentiate the best candidate. Our method, CODA, performs consensus-driven active model selection by modeling relationships between classifiers, categories, and data points within a probabilistic framework. The framework uses the consensus and disagreement between models in the candidate pool to guide the label acquisition process, and Bayesian inference to update beliefs about which model is best as more information is collected. We validate our approach by curating a collection of 26 benchmark tasks capturing a range of model selection scenarios. CODA outperforms existing methods for active model selection significantly, reducing the annotation effort required to discover the best model by upwards of 70% compared to the previous state-of-the-art. Code and data are available at https://github.com/justinkay/coda.

Oindrila Saha, Logan Lawrence, Grant Van Horn et al.

Transductive zero-shot learning with vision-language models leverages image-image similarities within the dataset to achieve better classification accuracy compared to the inductive setting. However, there is little work that explores the structure of the language space in this context. We propose GTA-CLIP, a novel technique that incorporates supervision from language models for joint transduction in language and vision spaces. Our approach is iterative and consists of three steps: (i) incrementally exploring the attribute space by querying language models, (ii) an attribute-augmented transductive inference procedure, and (iii) fine-tuning the language and vision encoders based on inferred labels within the dataset. Through experiments with CLIP encoders, we demonstrate that GTA-CLIP, yields an average performance improvement of 8.6% and 3.7% across 12 datasets and 3 encoders, over CLIP and transductive CLIP respectively in the zero-shot setting. We also observe similar improvements in a few-shot setting. We present ablation studies that demonstrate the value of each step and visualize how the vision and language spaces evolve over iterations driven by the transductive learning. Code is released at https://github.com/cvl-umass/GTA-CLIP

Aaron Sun, Subhransu Maji, Grant Van Horn

In the single-positive multi-label (SPML) setting, each image in a dataset is labeled with the presence of a single class, while the true presence of other classes remains unknown. The challenge is to narrow the performance gap between this partially-labeled setting and fully-supervised learning, which often requires a significant annotation budget. Prior SPML methods were developed and benchmarked on synthetic datasets created by randomly sampling single positive labels from fully-annotated datasets like Pascal VOC, COCO, NUS-WIDE, and CUB200. However, this synthetic approach does not reflect real-world scenarios and fails to capture the fine-grained complexities that can lead to difficult misclassifications. In this work, we introduce the L48 dataset, a fine-grained, real-world multi-label dataset derived from recordings of bird sounds. L48 provides a natural SPML setting with single-positive annotations on a challenging, fine-grained domain, as well as two extended settings in which domain priors give access to additional negative labels. We benchmark existing SPML methods on L48 and observe significant performance differences compared to synthetic datasets and analyze method weaknesses, underscoring the need for more realistic and difficult benchmarks.

Zenglin Shi, Pascal Mettes, Subhransu Maji et al.

The deep image prior showed that a randomly initialized network with a suitable architecture can be trained to solve inverse imaging problems by simply optimizing it's parameters to reconstruct a single degraded image. However, it suffers from two practical limitations. First, it remains unclear how to control the prior beyond the choice of the network architecture. Second, training requires an oracle stopping criterion as during the optimization the performance degrades after reaching an optimum value. To address these challenges we introduce a frequency-band correspondence measure to characterize the spectral bias of the deep image prior, where low-frequency image signals are learned faster and better than high-frequency counterparts. Based on our observations, we propose techniques to prevent the eventual performance degradation and accelerate convergence. We introduce a Lipschitz-controlled convolution layer and a Gaussian-controlled upsampling layer as plug-in replacements for layers used in the deep architectures. The experiments show that with these changes the performance does not degrade during optimization, relieving us from the need for an oracle stopping criterion. We further outline a stopping criterion to avoid superfluous computation. Finally, we show that our approach obtains favorable results compared to current approaches across various denoising, deblocking, inpainting, super-resolution and detail enhancement tasks. Code is available at \url{https://github.com/shizenglin/Measure-and-Control-Spectral-Bias}.

Jong-Chyi Su, Subhransu Maji

Semi-iNat is a challenging dataset for semi-supervised classification with a long-tailed distribution of classes, fine-grained categories, and domain shifts between labeled and unlabeled data. This dataset is behind the second iteration of the semi-supervised recognition challenge to be held at the FGVC8 workshop at CVPR 2021. Different from the previous one, this dataset (i) includes images of species from different kingdoms in the natural taxonomy, (ii) is at a larger scale -- with 810 in-class and 1629 out-of-class species for a total of 330k images, and (iii) does not provide in/out-of-class labels, but provides coarse taxonomic labels (kingdom and phylum) for the unlabeled images. This document describes baseline results and the details of the dataset which is available here: \url{https://github.com/cvl-umass/semi-inat-2021}.

Jong-Chyi Su, Subhransu Maji

This document describes the details and the motivation behind a new dataset we collected for the semi-supervised recognition challenge~\cite{semi-aves} at the FGVC7 workshop at CVPR 2020. The dataset contains 1000 species of birds sampled from the iNat-2018 dataset for a total of nearly 150k images. From this collection, we sample a subset of classes and their labels, while adding the images from the remaining classes to the unlabeled set of images. The presence of out-of-domain data (novel classes), high class-imbalance, and fine-grained similarity between classes poses significant challenges for existing semi-supervised recognition techniques in the literature. The dataset is available here: \url{https://github.com/cvl-umass/semi-inat-2020}

Zhaowei Cai, Avinash Ravichandran, Subhransu Maji et al.

We present a plug-in replacement for batch normalization (BN) called exponential moving average normalization (EMAN), which improves the performance of existing student-teacher based self- and semi-supervised learning techniques. Unlike the standard BN, where the statistics are computed within each batch, EMAN, used in the teacher, updates its statistics by exponential moving average from the BN statistics of the student. This design reduces the intrinsic cross-sample dependency of BN and enhances the generalization of the teacher. EMAN improves strong baselines for self-supervised learning by 4-6/1-2 points and semi-supervised learning by about 7/2 points, when 1%/10% supervised labels are available on ImageNet. These improvements are consistent across methods, network architectures, training duration, and datasets, demonstrating the general effectiveness of this technique. The code is available at https://github.com/amazon-research/exponential-moving-average-normalization.

Matheus Gadelha, Aruni RoyChowdhury, Gopal Sharma et al.

The problems of shape classification and part segmentation from 3D point clouds have garnered increasing attention in the last few years. Both of these problems, however, suffer from relatively small training sets, creating the need for statistically efficient methods to learn 3D shape representations. In this paper, we investigate the use of Approximate Convex Decompositions (ACD) as a self-supervisory signal for label-efficient learning of point cloud representations. We show that using ACD to approximate ground truth segmentation provides excellent self-supervision for learning 3D point cloud representations that are highly effective on downstream tasks. We report improvements over the state-of-the-art for unsupervised representation learning on the ModelNet40 shape classification dataset and significant gains in few-shot part segmentation on the ShapeNetPart dataset.Code available at https://github.com/matheusgadelha/PointCloudLearningACD

Kwonjoon Lee, Subhransu Maji, Avinash Ravichandran et al.

Many meta-learning approaches for few-shot learning rely on simple base learners such as nearest-neighbor classifiers. However, even in the few-shot regime, discriminatively trained linear predictors can offer better generalization. We propose to use these predictors as base learners to learn representations for few-shot learning and show they offer better tradeoffs between feature size and performance across a range of few-shot recognition benchmarks. Our objective is to learn feature embeddings that generalize well under a linear classification rule for novel categories. To efficiently solve the objective, we exploit two properties of linear classifiers: implicit differentiation of the optimality conditions of the convex problem and the dual formulation of the optimization problem. This allows us to use high-dimensional embeddings with improved generalization at a modest increase in computational overhead. Our approach, named MetaOptNet, achieves state-of-the-art performance on miniImageNet, tieredImageNet, CIFAR-FS, and FC100 few-shot learning benchmarks. Our code is available at https://github.com/kjunelee/MetaOptNet.

Mustafa Chasmai, Alexander Shepard, Subhransu Maji et al.

We present the iNaturalist Sounds Dataset (iNatSounds), a collection of 230,000 audio files capturing sounds from over 5,500 species, contributed by more than 27,000 recordists worldwide. The dataset encompasses sounds from birds, mammals, insects, reptiles, and amphibians, with audio and species labels derived from observations submitted to iNaturalist, a global citizen science platform. Each recording in the dataset varies in length and includes a single species annotation. We benchmark multiple backbone architectures, comparing multiclass classification objectives with multilabel objectives. Despite weak labeling, we demonstrate that iNatSounds serves as a useful pretraining resource by benchmarking it on strongly labeled downstream evaluation datasets. The dataset is available as a single, freely accessible archive, promoting accessibility and research in this important domain. We envision models trained on this data powering next-generation public engagement applications, and assisting biologists, ecologists, and land use managers in processing large audio collections, thereby contributing to the understanding of species compositions in diverse soundscapes.

Max Hamilton, Christian Lange, Elijah Cole et al.

Species range maps (SRMs) are essential tools for research and policy-making in ecology, conservation, and environmental management. However, traditional SRMs rely on the availability of environmental covariates and high-quality species location observation data, both of which can be challenging to obtain due to geographic inaccessibility and resource constraints. We propose a novel approach combining millions of citizen science species observations with textual descriptions from Wikipedia, covering habitat preferences and range descriptions for tens of thousands of species. Our framework maps locations, species, and text descriptions into a common space, facilitating the learning of rich spatial covariates at a global scale and enabling zero-shot range estimation from textual descriptions. Evaluated on held-out species, our zero-shot SRMs significantly outperform baselines and match the performance of SRMs obtained using tens of observations. Our approach also acts as a strong prior when combined with observational data, resulting in more accurate range estimation with less data. We present extensive quantitative and qualitative analyses of the learned representations in the context of range estimation and other spatial tasks, demonstrating the effectiveness of our approach.

Rangel Daroya, Elijah Cole, Oisin Mac Aodha et al.

Species distributions encode valuable ecological and environmental information, yet their potential for guiding representation learning in remote sensing remains underexplored. We introduce WildSAT, which pairs satellite images with millions of geo-tagged wildlife observations readily-available on citizen science platforms. WildSAT employs a contrastive learning approach that jointly leverages satellite images, species occurrence maps, and textual habitat descriptions to train or fine-tune models. This approach significantly improves performance on diverse satellite image recognition tasks, outperforming both ImageNet-pretrained models and satellite-specific baselines. Additionally, by aligning visual and textual information, WildSAT enables zero-shot retrieval, allowing users to search geographic locations based on textual descriptions. WildSAT surpasses recent cross-modal learning methods, including approaches that align satellite images with ground imagery or wildlife photos, demonstrating the advantages of our approach. Finally, we analyze the impact of key design choices and highlight the broad applicability of WildSAT to remote sensing and biodiversity monitoring.

Lukas Picek, Timm Haucke, Lukáš Adam et al.

Recognizing individual animals over time is central to many ecological and conservation questions, including estimating abundance, survival, movement, and social structure. Recent advances in automated identification from images and even acoustic data suggest that this process could be greatly accelerated, yet their promise has not translated well into ecological practice. We argue that the main barrier is not the performance of the automated methods themselves, but a mismatch between how those methods are typically developed and evaluated, and how ecological data is actually collected, processed, reviewed, and used. Future progress, therefore, will depend less on algorithmic gains alone than on recognizing that the usefulness of automated identification is grounded in ecological context: it depends on what question is being asked, what data are available, and what kinds of mistakes matter. Only by centering these questions can we move toward automated identification of individuals that is not only accurate but also ecologically useful, transparent, and trustworthy.

Gustavo Perez, Daniel Sheldon, Grant Van Horn et al.

Computer vision-based re-identification (Re-ID) systems are increasingly being deployed for estimating population size in large image collections. However, the estimated size can be significantly inaccurate when the task is challenging or when deployed on data from new distributions. We propose a human-in-the-loop approach for estimating population size driven by a pairwise similarity derived from an off-the-shelf Re-ID system. Our approach, based on nested importance sampling, selects pairs of images for human vetting driven by the pairwise similarity, and produces asymptotically unbiased population size estimates with associated confidence intervals. We perform experiments on various animal Re-ID datasets and demonstrate that our method outperforms strong baselines and active clustering approaches. In many cases, we are able to reduce the error rates of the estimated size from around 80% using CV alone to less than 20% by vetting a fraction (often less than 0.002%) of the total pairs. The cost of vetting reduces with the increase in accuracy and provides a practical approach for population size estimation within a desired tolerance when deploying Re-ID systems.

Christian Lange, Max Hamilton, Elijah Cole et al.

Knowing where a particular species can or cannot be found on Earth is crucial for ecological research and conservation efforts. By mapping the spatial ranges of all species, we would obtain deeper insights into how global biodiversity is affected by climate change and habitat loss. However, accurate range estimates are only available for a relatively small proportion of all known species. For the majority of the remaining species, we typically only have a small number of records denoting the spatial locations where they have previously been observed. We outline a new approach for few-shot species range estimation to address the challenge of accurately estimating the range of a species from limited data. During inference, our model takes a set of spatial locations as input, along with optional metadata such as text or an image, and outputs a species encoding that can be used to predict the range of a previously unseen species in a feedforward manner. We evaluate our approach on two challenging benchmarks, where we obtain state-of-the-art range estimation performance, in a fraction of the compute time, compared to recent alternative approaches.

Rangel Daroya, Luisa Vieira Lucchese, Travis Simmons et al.

Many remote sensing applications employ masking of pixels in satellite imagery for subsequent measurements. For example, estimating water quality variables, such as Suspended Sediment Concentration (SSC) requires isolating pixels depicting water bodies unaffected by clouds, their shadows, terrain shadows, and snow and ice formation. A significant bottleneck is the reliance on a variety of data products (e.g., satellite imagery, elevation maps), and a lack of precision in individual steps affecting estimation accuracy. We propose to improve both the accuracy and computational efficiency of masking by developing a system that predicts all required masks from Harmonized Landsat and Sentinel (HLS) imagery. Our model employs multi-tasking to share computation and enable higher accuracy across tasks. We experiment with recent advances in deep network architectures and show that masking models can benefit from these, especially when combined with pre-training on large satellite imagery datasets. We present a collection of models offering different speed/accuracy trade-offs for masking. MobileNet variants are the fastest, and perform competitively with larger architectures. Transformer-based architectures are the slowest, but benefit the most from pre-training on large satellite imagery datasets. Our models provide a 9% F1 score improvement compared to previous work on water pixel identification. When integrated with an SSC estimation system, our models result in a 30x speedup while reducing estimation error by 2.64 mg/L, allowing for global-scale analysis. We also evaluate our model on a recently proposed cloud and cloud shadow estimation benchmark, where we outperform the current state-of-the-art model by at least 6% in F1 score.

Rangel Daroya, Taylor Rowley, Jonathan Flores et al.

Surface water dynamics play a critical role in Earth's climate system, influencing ecosystems, agriculture, disaster resilience, and sustainable development. Yet monitoring rivers and surface water at fine spatial and temporal scales remains challenging -- especially for narrow or sediment-rich rivers that are poorly captured by low-resolution satellite data. To address this, we introduce RiverScope, a high-resolution dataset developed through collaboration between computer science and hydrology experts. RiverScope comprises 1,145 high-resolution images (covering 2,577 square kilometers) with expert-labeled river and surface water masks, requiring over 100 hours of manual annotation. Each image is co-registered with Sentinel-2, SWOT, and the SWOT River Database (SWORD), enabling the evaluation of cost-accuracy trade-offs across sensors -- a key consideration for operational water monitoring. We also establish the first global, high-resolution benchmark for river width estimation, achieving a median error of 7.2 meters -- significantly outperforming existing satellite-derived methods. We extensively evaluate deep networks across multiple architectures (e.g., CNNs and transformers), pretraining strategies (e.g., supervised and self-supervised), and training datasets (e.g., ImageNet and satellite imagery). Our best-performing models combine the benefits of transfer learning with the use of all the multispectral PlanetScope channels via learned adaptors. RiverScope provides a valuable resource for fine-scale and multi-sensor hydrological modeling, supporting climate adaptation and sustainable water management.

Max Hamilton, Jinlin Lai, Wenlong Zhao et al.

AI has the potential to transform scientific discovery by analyzing vast datasets with little human effort. However, current workflows often do not provide the accuracy or statistical guarantees that are needed. We introduce active measurement, a human-in-the-loop AI framework for scientific measurement. An AI model is used to predict measurements for individual units, which are then sampled for human labeling using importance sampling. With each new set of human labels, the AI model is improved and an unbiased Monte Carlo estimate of the total measurement is refined. Active measurement can provide precise estimates even with an imperfect AI model, and requires little human effort when the AI model is very accurate. We derive novel estimators, weighting schemes, and confidence intervals, and show that active measurement reduces estimation error compared to alternatives in several measurement tasks.

Logan Lawrence, Oindrila Saha, Megan Wei et al. · mit

Despite the renewed interest in zero-shot visual classification due to the rise of Multimodal Large Language Models (MLLMs), the problem of evaluating free-form responses of auto-regressive models remains a persistent challenge. Most existing works focus on language-only tasks or don't consider Multiple Choice Questions (MCQs) beyond 5-way options, both of which are critical capabilities to solve tasks in Fine-Grained Visual Classification (FGVC) where choice counts are in the hundreds to thousands and the choices are highly related. Furthermore, in this highly multi-way MCQ setting it is not clear how to extend LLM choice extraction to retrieval-based problems, where computing probabilities over the choice set is computationally costly. In this work we investigate nlg2choice, a simple two-stage method which first asks the MLLM an open-ended question for the task with minimal constraints, then uses text-only constrained decoding to predict the most likely choice. In retrieval settings, we compute the probability of the constrained response taking that choice with an early stopping method to significantly improve throughput. Our results show improvement over a suite of seven fine-grained visual datasets when evaluating in terms of classification and retrieval, and show that this performance holds over the various ways that users of LLMs can implement tasks in natural language.

Oindrila Saha, Vojtech Krs, Radomir Mech et al.

We present SIGMA-GEN, a unified framework for multi-identity preserving image generation. Unlike prior approaches, SIGMA-GEN is the first to enable single-pass multi-subject identity-preserved generation guided by both structural and spatial constraints. A key strength of our method is its ability to support user guidance at various levels of precision -- from coarse 2D or 3D boxes to pixel-level segmentations and depth -- with a single model. To enable this, we introduce SIGMA-SET27K, a novel synthetic dataset that provides identity, structure, and spatial information for over 100k unique subjects across 27k images. Through extensive evaluation we demonstrate that SIGMA-GEN achieves state-of-the-art performance in identity preservation, image generation quality, and speed. Code and visualizations at https://oindrilasaha.github.io/SIGMA-Gen/

Mustafa Chasmai, Gauri Jagatap, Gouthaman KV et al.

Recent advancements in video large language models (Video LLMs) have significantly advanced the field of video question answering (VideoQA). While existing methods perform well on short videos, they often struggle with long-range reasoning in longer videos. To scale Video LLMs for longer video content, frame sub-sampling (selecting frames at regular intervals) is commonly used. However, this approach is suboptimal, often leading to the loss of crucial frames or the inclusion of redundant information from multiple similar frames. Missing key frames impairs the model's ability to answer questions accurately, while redundant frames lead the model to focus on irrelevant video segments and increase computational resource consumption. In this paper, we investigate the use of a general-purpose text-to-video moment retrieval model to guide the frame sampling process. We propose "moment sampling", a novel, model-agnostic approach that enables the model to select the most relevant frames according to the context of the question. Specifically, we employ a lightweight moment retrieval model to prioritize frame selection. By focusing on the frames most pertinent to the given question, our method enhances long-form VideoQA performance in Video LLMs. Through extensive experiments on four long-form VideoQA datasets, using four state-of-the-art Video LLMs, we demonstrate the effectiveness of the proposed approach.

Mustafa Chasmai, Wuao Liu, Subhransu Maji et al.

Can we determine someone's geographic location purely from the sounds they hear? Are acoustic signals enough to localize within a country, state, or even city? We tackle the challenge of global-scale audio geolocation, formalize the problem, and conduct an in-depth analysis with wildlife audio from the iNatSounds dataset. Adopting a vision-inspired approach, we convert audio recordings to spectrograms and benchmark existing image geolocation techniques. We hypothesize that species vocalizations offer strong geolocation cues due to their defined geographic ranges and propose an approach that integrates species range prediction with retrieval-based geolocation. We further evaluate whether geolocation improves when analyzing species-rich recordings or when aggregating across spatiotemporal neighborhoods. Finally, we introduce case studies from movies to explore multimodal geolocation using both audio and visual content. Our work highlights the advantages of integrating audio and visual cues, and sets the stage for future research in audio geolocation.

Rangel Daroya, Aaron Sun, Subhransu Maji

Modeling and visualizing relationships between tasks or datasets is an important step towards solving various meta-tasks such as dataset discovery, multi-tasking, and transfer learning. However, many relationships, such as containment and transferability, are naturally asymmetric and current approaches for representation and visualization (e.g., t-SNE) do not readily support this. We propose Task2Box, an approach to represent tasks using box embeddings -- axis-aligned hyperrectangles in low dimensional spaces -- that can capture asymmetric relationships between them through volumetric overlaps. We show that Task2Box accurately predicts unseen hierarchical relationships between nodes in ImageNet and iNaturalist datasets, as well as transferability between tasks in the Taskonomy benchmark. We also show that box embeddings estimated from task representations (e.g., CLIP, Task2Vec, or attribute based) can be used to predict relationships between unseen tasks more accurately than classifiers trained on the same representations, as well as handcrafted asymmetric distances (e.g., KL divergence). This suggests that low-dimensional box embeddings can effectively capture these task relationships and have the added advantage of being interpretable. We use the approach to visualize relationships among publicly available image classification datasets on popular dataset hosting platform called Hugging Face.

Gopal Sharma, Bidya Dash, Aruni RoyChowdhury et al.

We present PriFit, a semi-supervised approach for label-efficient learning of 3D point cloud segmentation networks. PriFit combines geometric primitive fitting with point-based representation learning. Its key idea is to learn point representations whose clustering reveals shape regions that can be approximated well by basic geometric primitives, such as cuboids and ellipsoids. The learned point representations can then be re-used in existing network architectures for 3D point cloud segmentation, and improves their performance in the few-shot setting. According to our experiments on the widely used ShapeNet and PartNet benchmarks, PriFit outperforms several state-of-the-art methods in this setting, suggesting that decomposability into primitives is a useful prior for learning representations predictive of semantic parts. We present a number of ablative experiments varying the choice of geometric primitives and downstream tasks to demonstrate the effectiveness of the method.

Oindrila Saha, Zezhou Cheng, Subhransu Maji

Advances in generative modeling based on GANs has motivated the community to find their use beyond image generation and editing tasks. In particular, several recent works have shown that GAN representations can be re-purposed for discriminative tasks such as part segmentation, especially when training data is limited. But how do these improvements stack-up against recent advances in self-supervised learning? Motivated by this we present an alternative approach based on contrastive learning and compare their performance on standard few-shot part segmentation benchmarks. Our experiments reveal that not only do the GAN-based approach offer no significant performance advantage, their multi-step training is complex, nearly an order-of-magnitude slower, and can introduce additional bias. These experiments suggest that the inductive biases of generative models, such as their ability to disentangle shape and texture, are well captured by standard feed-forward networks trained using contrastive learning. These experiments suggest that the inductive biases present in current generative models, such as their ability to disentangle shape and texture, are well captured by standard feed-forward networks trained using contrastive learning.

Jong-Chyi Su, Subhransu Maji

We propose techniques to incorporate coarse taxonomic labels to train image classifiers in fine-grained domains. Such labels can often be obtained with a smaller effort for fine-grained domains such as the natural world where categories are organized according to a biological taxonomy. On the Semi-iNat dataset consisting of 810 species across three Kingdoms, incorporating Phylum labels improves the Species level classification accuracy by 6% in a transfer learning setting using ImageNet pre-trained models. Incorporating the hierarchical label structure with a state-of-the-art semi-supervised learning algorithm called FixMatch improves the performance further by 1.3%. The relative gains are larger when detailed labels such as Class or Order are provided, or when models are trained from scratch. However, we find that most methods are not robust to the presence of out-of-domain data from novel classes. We propose a technique to select relevant data from a large collection of unlabeled images guided by the hierarchy which improves the robustness. Overall, our experiments show that semi-supervised learning with coarse taxonomic labels are practical for training classifiers in fine-grained domains.

Gustavo Perez, Subhransu Maji

We consider the problem of adapting a network trained on three-channel color images to a hyperspectral domain with a large number of channels. To this end, we propose domain adaptor networks that map the input to be compatible with a network trained on large-scale color image datasets such as ImageNet. Adaptors enable learning on small hyperspectral datasets where training a network from scratch may not be effective. We investigate architectures and strategies for training adaptors and evaluate them on a benchmark consisting of multiple hyperspectral datasets. We find that simple schemes such as linear projection or subset selection are often the most effective, but can lead to a loss in performance in some cases. We also propose a novel multi-view adaptor where of the inputs are combined in an intermediate layer of the network in an order invariant manner that provides further improvements. We present extensive experiments by varying the number of training examples in the benchmark to characterize the accuracy and computational trade-offs offered by these adaptors.

Jong-Chyi Su, Zezhou Cheng, Subhransu Maji

We evaluate the effectiveness of semi-supervised learning (SSL) on a realistic benchmark where data exhibits considerable class imbalance and contains images from novel classes. Our benchmark consists of two fine-grained classification datasets obtained by sampling classes from the Aves and Fungi taxonomy. We find that recently proposed SSL methods provide significant benefits, and can effectively use out-of-class data to improve performance when deep networks are trained from scratch. Yet their performance pales in comparison to a transfer learning baseline, an alternative approach for learning from a few examples. Furthermore, in the transfer setting, while existing SSL methods provide improvements, the presence of out-of-class is often detrimental. In this setting, standard fine-tuning followed by distillation-based self-training is the most robust. Our work suggests that semi-supervised learning with experts on realistic datasets may require different strategies than those currently prevalent in the literature.

Orchid Majumder, Avinash Ravichandran, Subhransu Maji et al.

Few-shot learning aims to transfer information from one task to enable generalization on novel tasks given a few examples. This information is present both in the domain and the class labels. In this work we investigate the complementary roles of these two sources of information by combining instance-discriminative contrastive learning and supervised learning in a single framework called Supervised Momentum Contrastive learning (SUPMOCO). Our approach avoids a problem observed in supervised learning where information in images not relevant to the task is discarded, which hampers their generalization to novel tasks. We show that (self-supervised) contrastive learning and supervised learning are mutually beneficial, leading to a new state-of-the-art on the META-DATASET - a recently introduced benchmark for few-shot learning. Our method is based on a simple modification of MOCO and scales better than prior work on combining supervised and self-supervised learning. This allows us to easily combine data from multiple domains leading to further improvements.

Gustavo Perez, Matteo Messa, Daniela Calzetti et al.

We present a machine learning (ML) pipeline to identify star clusters in the multi{color images of nearby galaxies, from observations obtained with the Hubble Space Telescope as part of the Treasury Project LEGUS (Legacy ExtraGalactic Ultraviolet Survey). StarcNet (STAR Cluster classification NETwork) is a multi-scale convolutional neural network (CNN) which achieves an accuracy of 68.6% (4 classes)/86.0% (2 classes: cluster/non-cluster) for star cluster classification in the images of the LEGUS galaxies, nearly matching human expert performance. We test the performance of StarcNet by applying pre-trained CNN model to galaxies not included in the training set, finding accuracies similar to the reference one. We test the effect of StarcNet predictions on the inferred cluster properties by comparing multi-color luminosity functions and mass-age plots from catalogs produced by StarcNet and by human-labeling; distributions in luminosity, color, and physical characteristics of star clusters are similar for the human and ML classified samples. There are two advantages to the ML approach: (1) reproducibility of the classifications: the ML algorithm's biases are fixed and can be measured for subsequent analysis; and (2) speed of classification: the algorithm requires minutes for tasks that humans require weeks to months to perform. By achieving comparable accuracy to human classifiers, StarcNet will enable extending classifications to a larger number of candidate samples than currently available, thus increasing significantly the statistics for cluster studies.

Zitian Chen, Subhransu Maji, Erik Learned-Miller

Few-shot learning aims to build classifiers for new classes from a small number of labeled examples and is commonly facilitated by access to examples from a distinct set of 'base classes'. The difference in data distribution between the test set (novel classes) and the base classes used to learn an inductive bias often results in poor generalization on the novel classes. To alleviate problems caused by the distribution shift, previous research has explored the use of unlabeled examples from the novel classes, in addition to labeled examples of the base classes, which is known as the transductive setting. In this work, we show that, surprisingly, off-the-shelf self-supervised learning outperforms transductive few-shot methods by 3.9% for 5-shot accuracy on miniImageNet without using any base class labels. This motivates us to examine more carefully the role of features learned through self-supervision in few-shot learning. Comprehensive experiments are conducted to compare the transferability, robustness, efficiency, and the complementarity of supervised and self-supervised features.

Chenyun Wu, Zhe Lin, Scott Cohen et al.

We consider the problem of segmenting image regions given a natural language phrase, and study it on a novel dataset of 77,262 images and 345,486 phrase-region pairs. Our dataset is collected on top of the Visual Genome dataset and uses the existing annotations to generate a challenging set of referring phrases for which the corresponding regions are manually annotated. Phrases in our dataset correspond to multiple regions and describe a large number of object and stuff categories as well as their attributes such as color, shape, parts, and relationships with other entities in the image. Our experiments show that the scale and diversity of concepts in our dataset poses significant challenges to the existing state-of-the-art. We systematically handle the long-tail nature of these concepts and present a modular approach to combine category, attribute, and relationship cues that outperforms existing approaches.

Chenyun Wu, Mikayla Timm, Subhransu Maji

Textures in natural images can be characterized by color, shape, periodicity of elements within them, and other attributes that can be described using natural language. In this paper, we study the problem of describing visual attributes of texture on a novel dataset containing rich descriptions of textures, and conduct a systematic study of current generative and discriminative models for grounding language to images on this dataset. We find that while these models capture some properties of texture, they fail to capture several compositional properties, such as the colors of dots. We provide critical analysis of existing models by generating synthetic but realistic textures with different descriptions. Our dataset also allows us to train interpretable models and generate language-based explanations of what discriminative features are learned by deep networks for fine-grained categorization where texture plays a key role. We present visualizations of several fine-grained domains and show that texture attributes learned on our dataset offer improvements over expert-designed attributes on the Caltech-UCSD Birds dataset.

Zezhou Cheng, Jong-Chyi Su, Subhransu Maji

Given a collection of images, humans are able to discover landmarks by modeling the shared geometric structure across instances. This idea of geometric equivariance has been widely used for the unsupervised discovery of object landmark representations. In this paper, we develop a simple and effective approach by combining instance-discriminative and spatially-discriminative contrastive learning. We show that when a deep network is trained to be invariant to geometric and photometric transformations, representations emerge from its intermediate layers that are highly predictive of object landmarks. Stacking these across layers in a "hypercolumn" and projecting them using spatially-contrastive learning further improves their performance on matching and few-shot landmark regression tasks. We also present a unified view of existing equivariant and invariant representation learning approaches through the lens of contrastive learning, shedding light on the nature of invariances learned. Experiments on standard benchmarks for landmark learning, as well as a new challenging one we propose, show that the proposed approach surpasses prior state-of-the-art.