Aligning Forest and Trees in Images and Long Captions for Visually Grounded Understanding

Large vision-language models such as CLIP struggle with long captions because they align images and texts as undifferentiated wholes. Fine-grained vision-language understanding requires hierarchical semantics capturing both global context and localized details across visual and textual domains. Yet linguistic hierarchies from syntax or semantics rarely match visual organization, and purely visual hierarchies tend to fragment scenes into appearance-driven parts without semantic focus. We propose CAFT (Cross-domain Alignment of Forests and Trees), a hierarchical image-text representation learning framework that aligns global and local semantics across images and long captions without pixel-level supervision. Coupling a fine-to-coarse visual encoder with a hierarchical text transformer, it uses a hierarchical alignment loss that matches whole images with whole captions while biasing region-sentence correspondences, so that coarse semantics are built from fine-grained evidence rather than from aggregation untethered to part-level grounding. Trained on 30M image-text pairs, CAFT achieves state-of-the-art performance on six long-text retrieval benchmarks and exhibits strong scaling behavior. Experiments show that hierarchical cross-domain alignment enables fine-grained, visually grounded image-text representations to emerge without explicit region-level supervision.

SHED Light on Segmentation for Dense Prediction

Dense prediction infers per-pixel values from a single image and is fundamental to 3D perception and robotics. Although real-world scenes exhibit strong structure, existing methods treat it as an independent pixel-wise prediction, often resulting in structural inconsistencies. We propose SHED, a novel encoder-decoder architecture that enforces geometric prior explicitly by incorporating segmentation into dense prediction. By bidirectional hierarchical reasoning, segment tokens are hierarchically pooled in the encoder and unpooled in the decoder to reverse the hierarchy. The model is supervised only at the final output, allowing the segment hierarchy to emerge without explicit segmentation supervision. SHED improves depth boundary sharpness and segment coherence, while demonstrating strong cross-domain generalization from synthetic to the real-world environments. Its hierarchy-aware decoder better captures global 3D scene layouts, leading to improved semantic segmentation performance. Moreover, SHED enhances 3D reconstruction quality and reveals interpretable part-level structures that are often missed by conventional pixel-wise methods.

Next-Embedding Prediction Makes Strong Vision Learners

Inspired by the success of generative pretraining in natural language, we ask whether the same principles can yield strong self-supervised visual learners. Instead of training models to output features for downstream use, we train them to generate embeddings to perform predictive tasks directly. This work explores such a shift from learning representations to learning models. Specifically, models learn to predict future patch embeddings conditioned on past ones, using causal masking and stop gradient, which we refer to as Next-Embedding Predictive Autoregression (NEPA). We demonstrate that a simple Transformer pretrained on ImageNet-1k with next embedding prediction as its sole learning objective is effective - no pixel reconstruction, discrete tokens, contrastive loss, or task-specific heads. This formulation retains architectural simplicity and scalability, without requiring additional design complexity. NEPA achieves strong results across tasks, attaining 83.8% and 85.3% top-1 accuracy on ImageNet-1K with ViT-B and ViT-L backbones after fine-tuning, and transferring effectively to semantic segmentation on ADE20K. We believe generative pretraining from embeddings provides a simple, scalable, and potentially modality-agnostic alternative to visual self-supervised learning.

Open Ad-hoc Categorization with Contextualized Feature Learning

Adaptive categorization of visual scenes is essential for AI agents to handle changing tasks. Unlike fixed common categories for plants or animals, ad-hoc categories are created dynamically to serve specific goals. We study open ad-hoc categorization: Given a few labeled exemplars and abundant unlabeled data, the goal is to discover the underlying context and to expand ad-hoc categories through semantic extension and visual clustering around it. Building on the insight that ad-hoc and common categories rely on similar perceptual mechanisms, we propose OAK, a simple model that introduces a small set of learnable context tokens at the input of a frozen CLIP and optimizes with both CLIP's image-text alignment objective and GCD's visual clustering objective. On Stanford and Clevr-4 datasets, OAK achieves state-of-the-art in accuracy and concept discovery across multiple categorizations, including 87.4% novel accuracy on Stanford Mood, surpassing CLIP and GCD by over 50%. Moreover, OAK produces interpretable saliency maps, focusing on hands for Action, faces for Mood, and backgrounds for Location, promoting transparency and trust while enabling adaptive and generalizable categorization.

Unified Humanoid Fall-Safety Policy from a Few Demonstrations

Falling is an inherent risk of humanoid mobility. Maintaining stability is thus a primary safety focus in robot control and learning, yet no existing approach fully averts loss of balance. When instability does occur, prior work addresses only isolated aspects of falling: avoiding falls, choreographing a controlled descent, or standing up afterward. Consequently, humanoid robots lack integrated strategies for impact mitigation and prompt recovery when real falls defy these scripts. We aim to go beyond keeping balance to make the entire fall-and-recovery process safe and autonomous: prevent falls when possible, reduce impact when unavoidable, and stand up when fallen. By fusing sparse human demonstrations with reinforcement learning and an adaptive diffusion-based memory of safe reactions, we learn adaptive whole-body behaviors that unify fall prevention, impact mitigation, and rapid recovery in one policy. Experiments in simulation and on a Unitree G1 demonstrate robust sim-to-real transfer, lower impact forces, and consistently fast recovery across diverse disturbances, pointing towards safer, more resilient humanoids in real environments. Videos are available at https://firm2025.github.io/.

Normalize Filters! Classical Wisdom for Deep Vision

Classical image filters, such as those for averaging or differencing, are carefully normalized to ensure consistency, interpretability, and to avoid artifacts like intensity shifts, halos, or ringing. In contrast, convolutional filters learned end-to-end in deep networks lack such constraints. Although they may resemble wavelets and blob/edge detectors, they are not normalized in the same or any way. Consequently, when images undergo atmospheric transfer, their responses become distorted, leading to incorrect outcomes. We address this limitation by proposing filter normalization, followed by learnable scaling and shifting, akin to batch normalization. This simple yet effective modification ensures that the filters are atmosphere-equivariant, enabling co-domain symmetry. By integrating classical filtering principles into deep learning (applicable to both convolutional neural networks and convolution-dependent vision transformers), our method achieves significant improvements on artificial and natural intensity variation benchmarks. Our ResNet34 could even outperform CLIP by a large margin. Our analysis reveals that unnormalized filters degrade performance, whereas filter normalization regularizes learning, promotes diversity, and improves robustness and generalization.

Let Humanoids Hike! Integrative Skill Development on Complex Trails

Hiking on complex trails demands balance, agility, and adaptive decision-making over unpredictable terrain. Current humanoid research remains fragmented and inadequate for hiking: locomotion focuses on motor skills without long-term goals or situational awareness, while semantic navigation overlooks real-world embodiment and local terrain variability. We propose training humanoids to hike on complex trails, driving integrative skill development across visual perception, decision making, and motor execution. We develop a learning framework, LEGO-H, that enables a vision-equipped humanoid robot to hike complex trails autonomously. We introduce two technical innovations: 1) A temporal vision transformer variant - tailored into Hierarchical Reinforcement Learning framework - anticipates future local goals to guide movement, seamlessly integrating locomotion with goal-directed navigation. 2) Latent representations of joint movement patterns, combined with hierarchical metric learning - enhance Privileged Learning scheme - enable smooth policy transfer from privileged training to onboard execution. These components allow LEGO-H to handle diverse physical and environmental challenges without relying on predefined motion patterns. Experiments across varied simulated trails and robot morphologies highlight LEGO-H's versatility and robustness, positioning hiking as a compelling testbed for embodied autonomy and LEGO-H as a baseline for future humanoid development.

Learning Hierarchical Semantic Classification by Grounding on Consistent Image Segmentations

Hierarchical semantic classification requires the prediction of a taxonomy tree instead of a single flat level of the tree, where both accuracies at individual levels and consistency across levels matter. We can train classifiers for individual levels, which has accuracy but not consistency, or we can train only the finest level classification and infer higher levels, which has consistency but not accuracy. Our key insight is that hierarchical recognition should not be treated as multi-task classification, as each level is essentially a different task and they would have to compromise with each other, but be grounded on image segmentations that are consistent across semantic granularities. Consistency can in fact improve accuracy. We build upon recent work on learning hierarchical segmentation for flat-level recognition, and extend it to hierarchical recognition. It naturally captures the intuition that fine-grained recognition requires fine image segmentation whereas coarse-grained recognition requires coarse segmentation; they can all be integrated into one recognition model that drives fine-to-coarse internal visual parsing.Additionally, we introduce a Tree-path KL Divergence loss to enforce consistent accurate predictions across levels. Our extensive experimentation and analysis demonstrate our significant gains on predicting an accurate and consistent taxonomy tree.

Trajectory Regularization Enhances Self-Supervised Geometric Representation

Self-supervised learning (SSL) has proven effective in learning high-quality representations for various downstream tasks, with a primary focus on semantic tasks. However, its application in geometric tasks remains underexplored, partially due to the absence of a standardized evaluation method for geometric representations. To address this gap, we introduce a new pose-estimation benchmark for assessing SSL geometric representations, which demands training without semantic or pose labels and achieving proficiency in both semantic and geometric downstream tasks. On this benchmark, we study enhancing SSL geometric representations without sacrificing semantic classification accuracy. We find that leveraging mid-layer representations improves pose-estimation performance by 10-20%. Further, we introduce an unsupervised trajectory-regularization loss, which improves performance by an additional 4% and improves generalization ability on out-of-distribution data. We hope the proposed benchmark and methods offer new insights and improvements in self-supervised geometric representation learning.

Debiased Learning for Remote Sensing Data

Deep learning has had remarkable success at analyzing handheld imagery such as consumer photos due to the availability of large-scale human annotations (e.g., ImageNet). However, remote sensing data lacks such extensive annotation and thus potential for supervised learning. To address this, we propose a highly effective semi-supervised approach tailored specifically to remote sensing data. Our approach encompasses two key contributions. First, we adapt the FixMatch framework to remote sensing data by designing robust strong and weak augmentations suitable for this domain. Second, we develop an effective semi-supervised learning method by removing bias in imbalanced training data resulting from both actual labels and pseudo-labels predicted by the model. Our simple semi-supervised framework was validated by extensive experimentation. Using 30\% of labeled annotations, it delivers a 7.1\% accuracy gain over the supervised learning baseline and a 2.1\% gain over the supervised state-of-the-art CDS method on the remote sensing xView dataset.