arXiv subjects

Rajat Sahay

Publications and source records attributed to Rajat Sahay.

6 recordsLinked to original sources

arXiv2026

Orbis 2: A Hierarchical World Model for Driving

Current world models operate at a single level of abstraction, with most prioritizing perceptual fidelity while lacking the spatial reasoning and semantic understanding required for real-world downstream tasks. We present a hierarchical driving world model that factorizes future prediction across two levels operating at distinct temporal and abstraction scales: a high-level predictor that forecasts coarse scene structure over extended temporal horizons, and a low-level generator that produces detailed predictions conditioned on the high-level output. This decomposition yields high perceptual fidelity while also capturing strong spatial and semantic representations. We further show that pretraining with a diffusion forcing objective yields substantially richer internal representations than the standard teacher forcing objective, while teacher forcing -- predicting only the next frame from clean context -- produces more stable autoregressive rollouts. We therefore introduce a generic two-stage training paradigm that pretrains the model with diffusion forcing and fine-tunes with teacher forcing, combining the representational benefits of the former with the rollout stability of the latter. Our approach achieves state-of-the-art results across the standard suite of driving world model evaluations on established benchmarks, including long-horizon generation fidelity, steering responsiveness evaluated on counterfactual scenarios, and internal representation quality. Project page with code, demo, checkpoints and qualitative results: https://lmb-freiburg.github.io/orbis2.github.io/

cs.CV↗

arXiv2025

What Drives Compositional Generalization? The Importance of Continuous Training Objectives in Visual Generative Models

Compositional generalization, the ability to generate novel combinations of known concepts, is a key ingredient for visual generative models. Yet, not all mechanisms that enable or inhibit it are fully understood. In this work, we conduct a systematic study of how various design choices influence compositional generalization in image and video generation in a positive or negative way. Through controlled experiments, we identify two key factors: (i) whether the training objective operates on a discrete or continuous distribution, and (ii) to what extent conditioning provides information about the constituent concepts during training. Building on these insights, we show that relaxing the MaskGIT discrete loss with an auxiliary continuous JEPA-based objective can improve compositional performance in discrete models like MaskGIT.

cs.CV↗

arXiv2025

AILuminate: Introducing v1.0 of the AI Risk and Reliability Benchmark from MLCommons

The rapid advancement and deployment of AI systems have created an urgent need for standard safety-evaluation frameworks. This paper introduces AILuminate v1.0, the first comprehensive industry-standard benchmark for assessing AI-product risk and reliability. Its development employed an open process that included participants from multiple fields. The benchmark evaluates an AI system's resistance to prompts designed to elicit dangerous, illegal, or undesirable behavior in 12 hazard categories, including violent crimes, nonviolent crimes, sex-related crimes, child sexual exploitation, indiscriminate weapons, suicide and self-harm, intellectual property, privacy, defamation, hate, sexual content, and specialized advice (election, financial, health, legal). Our method incorporates a complete assessment standard, extensive prompt datasets, a novel evaluation framework, a grading and reporting system, and the technical as well as organizational infrastructure for long-term support and evolution. In particular, the benchmark employs an understandable five-tier grading scale (Poor to Excellent) and incorporates an innovative entropy-based system-response evaluation. In addition to unveiling the benchmark, this report also identifies limitations of our method and of building safety benchmarks generally, including evaluator uncertainty and the constraints of single-turn interactions. This work represents a crucial step toward establishing global standards for AI risk and reliability evaluation while acknowledging the need for continued development in areas such as multiturn interactions, multimodal understanding, coverage of additional languages, and emerging hazard categories. Our findings provide valuable insights for model developers, system integrators, and policymakers working to promote safer AI deployment.

cs.CY↗

arXiv2024

MoPEFT: A Mixture-of-PEFTs for the Segment Anything Model

The emergence of foundation models, such as the Segment Anything Model (SAM), has sparked interest in Parameter-Efficient Fine-Tuning (PEFT) methods that tailor these large models to application domains outside their training data. However, different PEFT techniques modify the representation of a model differently, making it a non-trivial task to select the most appropriate method for the domain of interest. We propose a new framework, Mixture-of-PEFTs methods (MoPEFT), that is inspired by traditional Mixture-of-Experts (MoE) methodologies and is utilized for fine-tuning SAM. Our MoPEFT framework incorporates three different PEFT techniques as submodules and dynamically learns to activate the ones that are best suited for a given data-task setup. We test our method on the Segment Anything Model and show that MoPEFT consistently outperforms other fine-tuning methods on the MESS benchmark.

cs.CV↗

arXiv2022

Dynamic Template Selection Through Change Detection for Adaptive Siamese Tracking

Deep Siamese trackers have recently gained much attention in recent years since they can track visual objects at high speeds. Additionally, adaptive tracking methods, where target samples collected by the tracker are employed for online learning, have achieved state-of-the-art accuracy. However, single object tracking (SOT) remains a challenging task in real-world application due to changes and deformations in a target object's appearance. Learning on all the collected samples may lead to catastrophic forgetting, and thereby corrupt the tracking model. In this paper, SOT is formulated as an online incremental learning problem. A new method is proposed for dynamic sample selection and memory replay, preventing template corruption. In particular, we propose a change detection mechanism to detect gradual changes in object appearance and select the corresponding samples for online adaption. In addition, an entropy-based sample selection strategy is introduced to maintain a diversified auxiliary buffer for memory replay. Our proposed method can be integrated into any object tracking algorithm that leverages online learning for model adaptation. Extensive experiments conducted on the OTB-100, LaSOT, UAV123, and TrackingNet datasets highlight the cost-effectiveness of our method, along with the contribution of its key components. Results indicate that integrating our proposed method into state-of-art adaptive Siamese trackers can increase the potential benefits of a template update strategy, and significantly improve performance.

cs.CV↗

arXiv2022

Generative Target Update for Adaptive Siamese Tracking

Siamese trackers perform similarity matching with templates (i.e., target models) to recursively localize objects within a search region. Several strategies have been proposed in the literature to update a template based on the tracker output, typically extracted from the target search region in the current frame, and thereby mitigate the effects of target drift. However, this may lead to corrupted templates, limiting the potential benefits of a template update strategy. This paper proposes a model adaptation method for Siamese trackers that uses a generative model to produce a synthetic template from the object search regions of several previous frames, rather than directly using the tracker output. Since the search region encompasses the target, attention from the search region is used for robust model adaptation. In particular, our approach relies on an auto-encoder trained through adversarial learning to detect changes in a target object's appearance and predict a future target template, using a set of target templates localized from tracker outputs at previous frames. To prevent template corruption during the update, the proposed tracker also performs change detection using the generative model to suspend updates until the tracker stabilizes, and robust matching can resume through dynamic template fusion. Extensive experiments conducted on VOT-16, VOT-17, OTB-50, and OTB-100 datasets highlight the effectiveness of our method, along with the impact of its key components. Results indicate that our proposed approach can outperform state-of-art trackers, and its overall robustness allows tracking for a longer time before failure.

cs.CV↗