Rohit Gupta

Xishun Liao, Xuanpeng Zhao, Ziran Wang et al.

Connected and automated vehicles (CAVs) are supposed to share the road with human-driven vehicles (HDVs) in a foreseeable future. Therefore, considering the mixed traffic environment is more pragmatic, as the well-planned operation of CAVs may be interrupted by HDVs. In the circumstance that human behaviors have significant impacts, CAVs need to understand HDV behaviors to make safe actions. In this study, we develop a Driver Digital Twin (DDT) for the online prediction of personalized lane change behavior, allowing CAVs to predict surrounding vehicles' behaviors with the help of the digital twin technology. DDT is deployed on a vehicle-edge-cloud architecture, where the cloud server models the driver behavior for each HDV based on the historical naturalistic driving data, while the edge server processes the real-time data from each driver with his/her digital twin on the cloud to predict the lane change maneuver. The proposed system is first evaluated on a human-in-the-loop co-simulation platform, and then in a field implementation with three passenger vehicles connected through the 4G/LTE cellular network. The lane change intention can be recognized in 6 seconds on average before the vehicle crosses the lane separation line, and the Mean Euclidean Distance between the predicted trajectory and GPS ground truth is 1.03 meters within a 4-second prediction window. Compared to the general model, using a personalized model can improve prediction accuracy by 27.8%. The demonstration video of the proposed system can be watched at https://youtu.be/5cbsabgIOdM.

Rohit Gupta, Naveed Akhtar, Ajmal Mian et al.

Contrastive self-supervised learning (CSL) has managed to match or surpass the performance of supervised learning in image and video classification. However, it is still largely unknown if the nature of the representations induced by the two learning paradigms is similar. We investigate this under the lens of adversarial robustness. Our analysis of the problem reveals that CSL has intrinsically higher sensitivity to perturbations over supervised learning. We identify the uniform distribution of data representation over a unit hypersphere in the CSL representation space as the key contributor to this phenomenon. We establish that this is a result of the presence of false negative pairs in the training process, which increases model sensitivity to input perturbations. Our finding is supported by extensive experiments for image and video classification using adversarial perturbations and other input corruptions. We devise a strategy to detect and remove false negative pairs that is simple, yet effective in improving model robustness with CSL training. We close up to 68% of the robustness gap between CSL and its supervised counterpart. Finally, we contribute to adversarial learning by incorporating our method in CSL. We demonstrate an average gain of about 5% over two different state-of-the-art methods in this domain.

Anthony Bloch, Leonardo Colombo, Rohit Gupta et al.

We investigate symmetry reduction of optimal control problems for left-invariant control systems on Lie groups, with partial symmetry breaking cost functions. Our approach emphasizes the role of variational principles and considers a discrete-time setting as well as the standard continuous-time formulation. Specifically, we recast the optimal control problem as a constrained variational problem with a partial symmetry breaking Lagrangian and obtain the Euler--Poincaré equations from a variational principle. By applying a Legendre transformation to it, we recover the Lie-Poisson equations obtained by A. D. Borum [Master's Thesis, University of Illinois at Urbana-Champaign, 2015] in the same context. We also discretize the variational principle in time and obtain the discrete-time Lie-Poisson equations. We illustrate the theory with some practical examples including a motion planning problem in the presence of an obstacle.

Rohit Gupta, Naveed Akhtar, Gaurav Kumar Nayak et al.

Black-box adversarial attacks present a realistic threat to action recognition systems. Existing black-box attacks follow either a query-based approach where an attack is optimized by querying the target model, or a transfer-based approach where attacks are generated using a substitute model. While these methods can achieve decent fooling rates, the former tends to be highly query-inefficient while the latter assumes extensive knowledge of the black-box model's training data. In this paper, we propose a new attack on action recognition that addresses these shortcomings by generating perturbations to disrupt the features learned by a pre-trained substitute model to reduce the number of queries. By using a nearly disjoint dataset to train the substitute model, our method removes the requirement that the substitute model be trained using the same dataset as the target model, and leverages queries to the target model to retain the fooling rate benefits provided by query-based methods. This ultimately results in attacks which are more transferable than conventional black-box attacks. Through extensive experiments, we demonstrate highly query-efficient black-box attacks with the proposed framework. Our method achieves 8% and 12% higher deception rates compared to state-of-the-art query-based and transfer-based attacks, respectively.

Yunsheng Ma, Amr Abdelraouf, Rohit Gupta et al.

Multimodal large language models (MLLMs) have demonstrated remarkable potential for enhancing scene understanding in autonomous driving systems through powerful logical reasoning capabilities. However, the deployment of these models faces significant challenges due to their substantial parameter sizes and computational demands, which often exceed the constraints of onboard computation. One major limitation arises from the large number of visual tokens required to capture fine-grained and long-context visual information, leading to increased latency and memory consumption. To address this issue, we propose Video Token Sparsification (VTS), a novel approach that leverages the inherent redundancy in consecutive video frames to significantly reduce the total number of visual tokens while preserving the most salient information. VTS employs a lightweight CNN-based proposal model to adaptively identify key frames and prune less informative tokens, effectively mitigating hallucinations and increasing inference throughput without compromising performance. We conduct comprehensive experiments on the DRAMA and LingoQA benchmarks, demonstrating the effectiveness of VTS in achieving up to a 33\% improvement in inference throughput and a 28\% reduction in memory usage compared to the baseline without compromising performance.

Rohit Gupta, Mamshad Nayeem Rizve, Jayakrishnan Unnikrishnan et al.

Pre-trained vision-language models (VLMs) have enabled significant progress in open vocabulary computer vision tasks such as image classification, object detection and image segmentation. Some recent works have focused on extending VLMs to open vocabulary single label action classification in videos. However, previous methods fall short in holistic video understanding which requires the ability to simultaneously recognize multiple actions and entities e.g., objects in the video in an open vocabulary setting. We formulate this problem as open vocabulary multilabel video classification and propose a method to adapt a pre-trained VLM such as CLIP to solve this task. We leverage large language models (LLMs) to provide semantic guidance to the VLM about class labels to improve its open vocabulary performance with two key contributions. First, we propose an end-to-end trainable architecture that learns to prompt an LLM to generate soft attributes for the CLIP text-encoder to enable it to recognize novel classes. Second, we integrate a temporal modeling module into CLIP's vision encoder to effectively model the spatio-temporal dynamics of video concepts as well as propose a novel regularized finetuning technique to ensure strong open vocabulary classification performance in the video domain. Our extensive experimentation showcases the efficacy of our approach on multiple benchmark datasets.

Amr Abdelraouf, Rohit Gupta, Kyungtae Han

Accurate prediction of vehicle trajectories is vital for advanced driver assistance systems and autonomous vehicles. Existing methods mainly rely on generic trajectory predictions derived from large datasets, overlooking the personalized driving patterns of individual drivers. To address this gap, we propose an approach for interaction-aware personalized vehicle trajectory prediction that incorporates temporal graph neural networks. Our method utilizes Graph Convolution Networks (GCN) and Long Short-Term Memory (LSTM) to model the spatio-temporal interactions between target vehicles and their surrounding traffic. To personalize the predictions, we establish a pipeline that leverages transfer learning: the model is initially pre-trained on a large-scale trajectory dataset and then fine-tuned for each driver using their specific driving data. We employ human-in-the-loop simulation to collect personalized naturalistic driving trajectories and corresponding surrounding vehicle trajectories. Experimental results demonstrate the superior performance of our personalized GCN-LSTM model, particularly for longer prediction horizons, compared to its generic counterpart. Moreover, the personalized model outperforms individual models created without pre-training, emphasizing the significance of pre-training on a large dataset to avoid overfitting. By incorporating personalization, our approach enhances trajectory prediction accuracy.

Yunsheng Ma, Can Cui, Xu Cao et al.

Autonomous driving (AD) has made significant strides in recent years. However, existing frameworks struggle to interpret and execute spontaneous user instructions, such as "overtake the car ahead." Large Language Models (LLMs) have demonstrated impressive reasoning capabilities showing potential to bridge this gap. In this paper, we present LaMPilot, a novel framework that integrates LLMs into AD systems, enabling them to follow user instructions by generating code that leverages established functional primitives. We also introduce LaMPilot-Bench, the first benchmark dataset specifically designed to quantitatively evaluate the efficacy of language model programs in AD. Adopting the LaMPilot framework, we conduct extensive experiments to assess the performance of off-the-shelf LLMs on LaMPilot-Bench. Our results demonstrate the potential of LLMs in handling diverse driving scenarios and following user instructions in driving. To facilitate further research in this area, we release our code and data at https://github.com/PurdueDigitalTwin/LaMPilot.

Jessica Echterhoff, An Yan, Kyungtae Han et al.

Concept bottleneck models have been successfully used for explainable machine learning by encoding information within the model with a set of human-defined concepts. In the context of human-assisted or autonomous driving, explainability models can help user acceptance and understanding of decisions made by the autonomous vehicle, which can be used to rationalize and explain driver or vehicle behavior. We propose a new approach using concept bottlenecks as visual features for control command predictions and explanations of user and vehicle behavior. We learn a human-understandable concept layer that we use to explain sequential driving scenes while learning vehicle control commands. This approach can then be used to determine whether a change in a preferred gap or steering commands from a human (or autonomous vehicle) is led by an external stimulus or change in preferences. We achieve competitive performance to latent visual features while gaining interpretability within our model setup.

Rohit Gupta, Anirban Roy, Claire Christensen et al.

The recent growth in the consumption of online media by children during early childhood necessitates data-driven tools enabling educators to filter out appropriate educational content for young learners. This paper presents an approach for detecting educational content in online videos. We focus on two widely used educational content classes: literacy and math. For each class, we choose prominent codes (sub-classes) based on the Common Core Standards. For example, literacy codes include `letter names', `letter sounds', and math codes include `counting', `sorting'. We pose this as a fine-grained multilabel classification problem as videos can contain multiple types of educational content and the content classes can get visually similar (e.g., `letter names' vs `letter sounds'). We propose a novel class prototypes based supervised contrastive learning approach that can handle fine-grained samples associated with multiple labels. We learn a class prototype for each class and a loss function is employed to minimize the distances between a class prototype and the samples from the class. Similarly, distances between a class prototype and the samples from other classes are maximized. As the alignment between visual and audio cues are crucial for effective comprehension, we consider a multimodal transformer network to capture the interaction between visual and audio cues in videos while learning the embedding for videos. For evaluation, we present a dataset, APPROVE, employing educational videos from YouTube labeled with fine-grained education classes by education researchers. APPROVE consists of 193 hours of expert-annotated videos with 19 classes. The proposed approach outperforms strong baselines on APPROVE and other benchmarks such as Youtube-8M, and COIN. The dataset is available at https://github.com/rohit-gupta/MMContrast/tree/main/APPROVE

Vishal Narnaware, Ashmal Vayani, Rohit Gupta et al.

Stereotype biases in Large Multimodal Models (LMMs) perpetuate harmful societal prejudices, undermining the fairness and equity of AI applications. As LMMs grow increasingly influential, addressing and mitigating inherent biases related to stereotypes, harmful generations, and ambiguous assumptions in real-world scenarios has become essential. However, existing datasets evaluating stereotype biases in LMMs often lack diversity and rely on synthetic images, leaving a gap in bias evaluation for real-world visual contexts. To address this, we introduce the Stereotype Bias Benchmark (SB-bench), the most comprehensive framework to date for assessing stereotype biases across nine diverse categories with non-synthetic images. SB-bench rigorously evaluates LMMs through carefully curated, visually grounded scenarios, challenging them to reason accurately about visual stereotypes. It offers a robust evaluation framework featuring real-world visual samples, image variations, and multiple-choice question formats. By introducing visually grounded queries that isolate visual biases from textual ones, SB-bench enables a precise and nuanced assessment of a model's reasoning capabilities across varying levels of difficulty. Through rigorous testing of state-of-the-art open-source and closed-source LMMs, SB-bench provides a systematic approach to assessing stereotype biases in LMMs across key social dimensions. This benchmark represents a significant step toward fostering fairness in AI systems and reducing harmful biases, laying the groundwork for more equitable and socially responsible LMMs. Our code and dataset are publicly available.

Sung-Yeon Park, Can Cui, Yunsheng Ma et al.

Recent advances in multi-modal large language models (MLLMs) have demonstrated strong performance across various domains; however, their ability to comprehend driving scenes remains less proven. The complexity of driving scenarios, which includes multi-view information, poses significant challenges for existing MLLMs. In this paper, we introduce NuPlanQA-Eval, a multi-view, multi-modal evaluation benchmark for driving scene understanding. To further support generalization to multi-view driving scenarios, we also propose NuPlanQA-1M, a large-scale dataset comprising 1M real-world visual question-answering (VQA) pairs. For context-aware analysis of traffic scenes, we categorize our dataset into nine subtasks across three core skills: Road Environment Perception, Spatial Relations Recognition, and Ego-Centric Reasoning. Furthermore, we present BEV-LLM, integrating Bird's-Eye-View (BEV) features from multi-view images into MLLMs. Our evaluation results reveal key challenges that existing MLLMs face in driving scene-specific perception and spatial reasoning from ego-centric perspectives. In contrast, BEV-LLM demonstrates remarkable adaptability to this domain, outperforming other models in six of the nine subtasks. These findings highlight how BEV integration enhances multi-view MLLMs while also identifying key areas that require further refinement for effective adaptation to driving scenes. To facilitate further research, we publicly release NuPlanQA at https://github.com/sungyeonparkk/NuPlanQA.

Parth Parag Kulkarni, Rohit Gupta, Prakash Chandra Chhipa et al.

The task of video geolocalization aims to determine the precise GPS coordinates of a video's origin and map its trajectory; with applications in forensics, social media, and exploration. Existing classification-based approaches operate at a coarse city-level granularity and fail to capture fine-grained details, while image retrieval methods are impractical on a global scale due to the need for extensive image galleries which are infeasible to compile. Comparatively, constructing a gallery of GPS coordinates is straightforward and inexpensive. We propose VidTAG, a dual-encoder framework that performs frame-to-GPS retrieval using both self-supervised and language-aligned features. To address temporal inconsistencies in video predictions, we introduce the TempGeo module, which aligns frame embeddings, and the GeoRefiner module, an encoder-decoder architecture that refines GPS features using the aligned frame embeddings. Evaluations on Mapillary (MSLS) and GAMa datasets demonstrate our model's ability to generate temporally consistent trajectories and outperform baselines, achieving a 20% improvement at the 1 km threshold over GeoCLIP. We also beat current State-of-the-Art by 25% on global coarse grained video geolocalization (CityGuessr68k). Our approach enables fine-grained video geolocalization and lays a strong foundation for future research. More details on the project webpage: https://parthpk.github.io/vidtag_webpage/

Sirnam Swetha, Rohit Gupta, Parth Parag Kulkarni et al.

Video Question Answering (VideoQA) has made significant strides by leveraging multimodal learning to align visual and textual modalities. However, current benchmarks overwhelmingly focus on questions answerable through explicit visual content - actions, objects, and events directly observable within individual frames or short clips. In contrast, creative and cinematic videos - such as movies, TV shows, and narrative-driven content - employ storytelling techniques that deliberately omit certain depictions, requiring viewers to infer motives, relationships across discontinuous frames with disjoint visual contexts. Humans naturally excel at such implicit reasoning, seamlessly integrating information across time and context to construct coherent narratives. Yet current benchmarks fail to capture this essential dimension of human-like understanding. To bridge this gap, we present ImplicitQA, a novel benchmark specifically designed to test VideoQA models on human-like implicit reasoning. ImplicitQA comprises 1K meticulously annotated QA pairs drawn from 1K high-quality creative video clips covering 15 genres across 7 decades of content. Questions are systematically categorized into nine key reasoning dimensions: lateral and vertical spatial reasoning, depth and proximity, viewpoint and visibility, motion and trajectory, causal and motivational reasoning, social interactions, physical context, and inferred counting. These annotations are deliberately challenging, crafted by authors, validated through multiple annotators, and benchmarked against human performance to ensure high quality. Our extensive evaluations on 11 leading VideoQA models reveals consistent and significant performance degradation, underscoring their reliance on surface-level visual cues and highlighting the difficulty of implicit reasoning. https://huggingface.co/datasets/ucf-crcv/ImplicitQA.

Rohit Gupta, Jayakrishnan Unnikrishnan, Fan Fei et al.

Video Large Language Models (VideoLLMs) excel at video understanding tasks where outputs are textual, such as Video Question Answering and Video Captioning. However, they underperform specialized embedding-based models in Retrieval tasks, such as Text-toVideo Retrieval and Moment Retrieval. We introduce ViLL-E (Video-LLM-Embed), a unified VideoLLM architecture endowed with a novel embedding generation mechanism that allows the model to "think longer" for complex videos and stop early for easy ones. We train this model with a three-stage training methodology combining generative and contrastive learning: initial large-scale pre-training with video-caption pairs; followed by continual training on a smaller, detailed-caption dataset; and concluding with task-specific fine-tuning on a novel multi-task dataset covering Video QA, Temporal Localization, Video Retrieval, and Video-Text Matching. Our model significantly improves temporal localization (on avg. 7% over other VideoLLMs) and video retrieval (up to 4% over dual encoder models), achieving performance comparable to state-of-the-art specialized embedding models while remaining competitive on VideoQA tasks. Furthermore, our joint contrastive-generative training unlocks new zero-shot capabilities, significantly outperforming state-of-the-art methods in composed video retrieval (+5% over SotA) and retrieval from long text (+2% over SotA).

Sabbir Mollah, Rohit Gupta, Sirnam Swetha et al.

Employing a single, unified model (UM) for both visual understanding (image-to-text: I2T) and visual generation (text-to-image: T2I) has opened a new direction in Visual Language Model (VLM) research. While UMs can also support broader unimodal tasks (e.g., text-to-text, image-to-image), we focus on the core cross-modal pair T2I and I2T. Existing evaluation benchmarks consider these capabilities in isolation: FID and GenEval for T2I, and benchmarks such as MME, MMBench for I2T. These isolated single-pass metrics do not reveal cross-consistency: whether a model that "understands" a concept can also "render" it, nor whether semantic meaning is preserved when cycling between image and text modalities. To address this, we introduce the Semantic Drift Protocol (SDP) for Unified Models, a cyclic evaluation protocol that alternates I2T and T2I over multiple generations to quantify semantic drift. We propose two metrics: (i) Mean Cumulative Drift (MCD), an embedding-based measure of overall semantic drift; and (ii) Multi-Generation GenEval (MGG), an object-level compliance score extending GenEval. To assess generalization beyond COCO dataset, which is widely used in training; we create a new benchmark Nocaps+Docci400, sampled from NoCaps and DOCCI and evaluated on seven recent models. SDP reveals substantial variation in cross-modal stability: some models like BAGEL maintain semantic meaning over many alternations, whereas others like VILA-U drift quickly despite strong single-pass scores. Our results highlight SDP as a necessary complement to standard I2T and T2I evaluations. Code is available at https://github.com/mollahsabbir/Semantic-Drift-in-Unified-Models

Ron Campos, Ashmal Vayani, Parth Parag Kulkarni et al.

Image geolocalization, in which an AI model traditionally predicts the precise GPS coordinates of an image, is a challenging task with many downstream applications. However, the user cannot utilize the model to further their knowledge beyond the GPS coordinates; the model lacks an understanding of the location and the conversational ability to communicate with the user. In recent days, with the tremendous progress of large multimodal models (LMMs) -- proprietary and open-source -- researchers have attempted to geolocalize images via LMMs. However, the issues remain unaddressed; beyond general tasks, for more specialized downstream tasks, such as geolocalization, LMMs struggle. In this work, we propose solving this problem by introducing a conversational model, GAEA, that provides information regarding the location of an image as the user requires. No large-scale dataset enabling the training of such a model exists. Thus, we propose GAEA-1.4M, a comprehensive dataset comprising over 800k images and approximately 1.4M question-answer pairs, constructed by leveraging OpenStreetMap (OSM) attributes and geographical context clues. For quantitative evaluation, we propose a diverse benchmark, GAEA-Bench, comprising 3.5k image-text pairs to evaluate conversational capabilities equipped with diverse question types. We consider 11 state-of-the-art open-source and proprietary LMMs and demonstrate that GAEA significantly outperforms the best open-source model, LLaVA-OneVision, by 18.2% and the best proprietary model, GPT-4o, by 7.2%. Our dataset, model and codes are available.

Yunsheng Ma, Liangqi Yuan, Amr Abdelraouf et al.

Ensuring traffic safety and preventing accidents is a critical goal in daily driving, where the advancement of computer vision technologies can be leveraged to achieve this goal. In this paper, we present a multi-view, multi-scale framework for naturalistic driving action recognition and localization in untrimmed videos, namely M$^2$DAR, with a particular focus on detecting distracted driving behaviors. Our system features a weight-sharing, multi-scale Transformer-based action recognition network that learns robust hierarchical representations. Furthermore, we propose a new election algorithm consisting of aggregation, filtering, merging, and selection processes to refine the preliminary results from the action recognition module across multiple views. Extensive experiments conducted on the 7th AI City Challenge Track 3 dataset demonstrate the effectiveness of our approach, where we achieved an overlap score of 0.5921 on the A2 test set. Our source code is available at \url{https://github.com/PurdueDigitalTwin/M2DAR}.

Ishan Dave, Rohit Gupta, Mamshad Nayeem Rizve et al.

Contrastive learning has nearly closed the gap between supervised and self-supervised learning of image representations, and has also been explored for videos. However, prior work on contrastive learning for video data has not explored the effect of explicitly encouraging the features to be distinct across the temporal dimension. We develop a new temporal contrastive learning framework consisting of two novel losses to improve upon existing contrastive self-supervised video representation learning methods. The local-local temporal contrastive loss adds the task of discriminating between non-overlapping clips from the same video, whereas the global-local temporal contrastive aims to discriminate between timesteps of the feature map of an input clip in order to increase the temporal diversity of the learned features. Our proposed temporal contrastive learning framework achieves significant improvement over the state-of-the-art results in various downstream video understanding tasks such as action recognition, limited-label action classification, and nearest-neighbor video retrieval on multiple video datasets and backbones. We also demonstrate significant improvement in fine-grained action classification for visually similar classes. With the commonly used 3D ResNet-18 architecture with UCF101 pretraining, we achieve 82.4\% (+5.1\% increase over the previous best) top-1 accuracy on UCF101 and 52.9\% (+5.4\% increase) on HMDB51 action classification, and 56.2\% (+11.7\% increase) Top-1 Recall on UCF101 nearest neighbor video retrieval. Code released at github.com/DAVEISHAN/TCLR.

Dominika Tkaczyk, Rohit Gupta, Riccardo Cinti et al.

Bibliographic reference parsers extract machine-readable metadata such as author names, title, journal, and year from bibliographic reference strings. To extract the metadata, the parsers apply heuristics or machine learning. However, no reference parser, and no algorithm, consistently gives the best results in every scenario. For instance, one tool may be best in extracting titles in ACM citation style, but only third best when APA is used. Another tool may be best in extracting English author names, while another one is best for noisy data (i.e. inconsistent citation styles). In this paper, which is an extended version of our recent RecSys poster, we address the problem of reference parsing from a recommender-systems and meta-learning perspective. We propose ParsRec, a meta-learning based recommender-system that recommends the potentially most effective parser for a given reference string. ParsRec recommends one out of 10 open-source parsers: Anystyle-Parser, Biblio, CERMINE, Citation, Citation-Parser, GROBID, ParsCit, PDFSSA4MET, Reference Tagger, and Science Parse. We evaluate ParsRec on 105k references from chemistry. We propose two approaches to meta-learning recommendations. The first approach learns the best parser for an entire reference string. The second approach learns the best parser for each metadata type in a reference string. The second approach achieved a 2.6% increase in F1 (0.909 vs. 0.886) over the best single parser (GROBID), reducing the false positive rate by 20.2% (0.075 vs. 0.094), and the false negative rate by 18.9% (0.107 vs. 0.132).

Raj Nath Patel, Rohit Gupta, Prakash B. Pimpale et al.

Reordering is a preprocessing stage for Statistical Machine Translation (SMT) system where the words of the source sentence are reordered as per the syntax of the target language. We are proposing a rich set of rules for better reordering. The idea is to facilitate the training process by better alignments and parallel phrase extraction for a phrase-based SMT system. Reordering also helps the decoding process and hence improving the machine translation quality. We have observed significant improvements in the translation quality by using our approach over the baseline SMT. We have used BLEU, NIST, multi-reference word error rate, multi-reference position independent error rate for judging the improvements. We have exploited open source SMT toolkit MOSES to develop the system.

Ashmal Vayani, Dinura Dissanayake, Hasindri Watawana et al. · mila

Existing Large Multimodal Models (LMMs) generally focus on only a few regions and languages. As LMMs continue to improve, it is increasingly important to ensure they understand cultural contexts, respect local sensitivities, and support low-resource languages, all while effectively integrating corresponding visual cues. In pursuit of culturally diverse global multimodal models, our proposed All Languages Matter Benchmark (ALM-bench) represents the largest and most comprehensive effort to date for evaluating LMMs across 100 languages. ALM-bench challenges existing models by testing their ability to understand and reason about culturally diverse images paired with text in various languages, including many low-resource languages traditionally underrepresented in LMM research. The benchmark offers a robust and nuanced evaluation framework featuring various question formats, including true/false, multiple choice, and open-ended questions, which are further divided into short and long-answer categories. ALM-bench design ensures a comprehensive assessment of a model's ability to handle varied levels of difficulty in visual and linguistic reasoning. To capture the rich tapestry of global cultures, ALM-bench carefully curates content from 13 distinct cultural aspects, ranging from traditions and rituals to famous personalities and celebrations. Through this, ALM-bench not only provides a rigorous testing ground for state-of-the-art open and closed-source LMMs but also highlights the importance of cultural and linguistic inclusivity, encouraging the development of models that can serve diverse global populations effectively. Our benchmark is publicly available.

Chuheng Wei, Guoyuan Wu, Matthew J. Barth et al.

Reliable prediction of vehicle trajectories at signalized intersections is crucial to urban traffic management and autonomous driving systems. However, it presents unique challenges, due to the complex roadway layout at intersections, involvement of traffic signal controls, and interactions among different types of road users. To address these issues, we present in this paper a novel model called Knowledge-Informed Generative Adversarial Network (KI-GAN), which integrates both traffic signal information and multi-vehicle interactions to predict vehicle trajectories accurately. Additionally, we propose a specialized attention pooling method that accounts for vehicle orientation and proximity at intersections. Based on the SinD dataset, our KI-GAN model is able to achieve an Average Displacement Error (ADE) of 0.05 and a Final Displacement Error (FDE) of 0.12 for a 6-second observation and 6-second prediction cycle. When the prediction window is extended to 9 seconds, the ADE and FDE values are further reduced to 0.11 and 0.26, respectively. These results demonstrate the effectiveness of the proposed KI-GAN model in vehicle trajectory prediction under complex scenarios at signalized intersections, which represents a significant advancement in the target field.

Junda Wu, Jessica Echterhoff, Kyungtae Han et al.

Understanding a driver's behavior and intentions is important for potential risk assessment and early accident prevention. Safety and driver assistance systems can be tailored to individual drivers' behavior, significantly enhancing their effectiveness. However, existing datasets are limited in describing and explaining general vehicle movements based on external visual evidence. This paper introduces a benchmark, PDB-Eval, for a detailed understanding of Personalized Driver Behavior, and aligning Large Multimodal Models (MLLMs) with driving comprehension and reasoning. Our benchmark consists of two main components, PDB-X and PDB-QA. PDB-X can evaluate MLLMs' understanding of temporal driving scenes. Our dataset is designed to find valid visual evidence from the external view to explain the driver's behavior from the internal view. To align MLLMs' reasoning abilities with driving tasks, we propose PDB-QA as a visual explanation question-answering task for MLLM instruction fine-tuning. As a generic learning task for generative models like MLLMs, PDB-QA can bridge the domain gap without harming MLLMs' generalizability. Our evaluation indicates that fine-tuning MLLMs on fine-grained descriptions and explanations can effectively bridge the gap between MLLMs and the driving domain, which improves zero-shot performance on question-answering tasks by up to 73.2%. We further evaluate the MLLMs fine-tuned on PDB-X in Brain4Cars' intention prediction and AIDE's recognition tasks. We observe up to 12.5% performance improvements on the turn intention prediction task in Brain4Cars, and consistent performance improvements up to 11.0% on all tasks in AIDE.

Chuheng Wei, Ziye Qin, Siyan Li et al.

Driving behavior is inherently personal, influenced by individual habits, decision-making styles, and physiological states. However, most existing datasets treat all drivers as homogeneous, overlooking driver-specific variability. To address this gap, we introduce the Personalized Driving Behavior (PDB) dataset, a multi-modal dataset designed to capture personalization in driving behavior under naturalistic driving conditions. Unlike conventional datasets, PDB minimizes external influences by maintaining consistent routes, vehicles, and lighting conditions across sessions. It includes sources from 128-line LiDAR, front-facing camera video, GNSS, 9-axis IMU, CAN bus data (throttle, brake, steering angle), and driver-specific signals such as facial video and heart rate. The dataset features 12 participants, approximately 270,000 LiDAR frames, 1.6 million images, and 6.6 TB of raw sensor data. The processed trajectory dataset consists of 1,669 segments, each spanning 10 seconds with a 0.2-second interval. By explicitly capturing drivers' behavior, PDB serves as a unique resource for human factor analysis, driver identification, and personalized mobility applications, contributing to the development of human-centric intelligent transportation systems.

Nyle Siddiqui, Rohit Gupta, Sirnam Swetha et al.

State space models (SSMs) have emerged as a competitive alternative to transformers in various tasks. Their linear complexity and hidden-state recurrence make them particularly attractive for modeling long sequences, whereas attention becomes quadratically expensive. However, current training methods for video understanding are tailored towards transformers and fail to fully leverage the unique attributes of SSMs. For example, video models are often trained at a fixed resolution and video length to balance the quadratic scaling of attention cost against performance. Consequently, these models suffer from degraded performance when evaluated on videos with spatial and temporal resolutions unseen during training; a property we call spatio-temporal inflexibility. In the context of action recognition, this severely limits a model's ability to retain performance across both short- and long-form videos. Therefore, we propose a flexible training method that leverages and improves the inherent adaptability of SSMs. Our method samples videos at varying temporal and spatial resolutions during training and dynamically interpolates model weights to accommodate any spatio-temporal scale. This instills our SSM, which we call StretchySnake, with spatio-temporal flexibility and enables it to seamlessly handle videos ranging from short, fine-grained clips to long, complex activities. We introduce and compare five different variants of flexible training, and identify the most effective strategy for video SSMs. On short-action (UCF-101, HMDB-51) and long-action (COIN, Breakfast) benchmarks, StretchySnake outperforms transformer and SSM baselines alike by up to 28%, with strong adaptability to fine-grained actions (SSV2, Diving-48). Therefore, our method provides a simple drop-in training recipe that makes video SSMs more robust, resolution-agnostic, and efficient across diverse action recognition scenarios.

Jyoti Kini, Rohit Gupta, Mubarak Shah

Traditional object detection models are typically trained on a fixed set of classes, limiting their flexibility and making it costly to incorporate new categories. Open-vocabulary object detection addresses this limitation by enabling models to identify unseen classes without explicit training. Leveraging pretrained models contrastively trained on abundantly available ground-view image-text classification pairs provides a strong foundation for open-vocabulary object detection in aerial imagery. Domain shifts, viewpoint variations, and extreme scale differences make direct knowledge transfer across domains ineffective, requiring specialized adaptation strategies. In this paper, we propose a novel framework for adapting open-vocabulary representations from ground-view images to solve object detection in aerial imagery through structured domain alignment. The method introduces contrastive image-to-image alignment to enhance the similarity between aerial and ground-view embeddings and employs multi-instance vocabulary associations to align aerial images with text embeddings. Extensive experiments on the xView, DOTAv2, VisDrone, DIOR, and HRRSD datasets are used to validate our approach. Our open-vocabulary model achieves improvements of +6.32 mAP on DOTAv2, +4.16 mAP on VisDrone (Images), and +3.46 mAP on HRRSD in the zero-shot setting when compared to finetuned closed-vocabulary dataset-specific model performance, thus paving the way for more flexible and scalable object detection systems in aerial applications.

Sung-Yeon Park, Adam Lee, Juanwu Lu et al.

Driving scene manipulation with sensor data is emerging as a promising alternative to traditional virtual driving simulators. However, existing frameworks struggle to generate realistic scenarios efficiently due to limited editing capabilities. To address these challenges, we present SIMSplat, a predictive driving scene editor with language-aligned Gaussian splatting. As a language-controlled editor, SIMSplat enables intuitive manipulation using natural language prompts. By aligning language with Gaussian-reconstructed scenes, it further supports direct querying of road objects, allowing precise and flexible editing. Our method provides detailed object-level editing, including adding new objects and modifying the trajectories of both vehicles and pedestrians, while also incorporating predictive path refinement through multi-agent motion prediction to generate realistic interactions among all agents in the scene. Experiments on the Waymo dataset demonstrate SIMSplat's extensive editing capabilities and adaptability across a wide range of scenarios. Project page: https://sungyeonparkk.github.io/simsplat/

Kyungtae Han, Yitao Chen, Rohit Gupta et al.

While autonomous driving technologies continue to advance, current Advanced Driver Assistance Systems (ADAS) remain limited in their ability to interpret scene context or engage with drivers through natural language. These systems typically rely on predefined logic and lack support for dialogue-based interaction, making them inflexible in dynamic environments or when adapting to driver intent. This paper presents Scene-Aware Conversational ADAS (SC-ADAS), a modular framework that integrates Generative AI components including large language models, vision-to-text interpretation, and structured function calling to enable real-time, interpretable, and adaptive driver assistance. SC-ADAS supports multi-turn dialogue grounded in visual and sensor context, allowing natural language recommendations and driver-confirmed ADAS control. Implemented in the CARLA simulator with cloud-based Generative AI, the system executes confirmed user intents as structured ADAS commands without requiring model fine-tuning. We evaluate SC-ADAS across scene-aware, conversational, and revisited multi-turn interactions, highlighting trade-offs such as increased latency from vision-based context retrieval and token growth from accumulated dialogue history. These results demonstrate the feasibility of combining conversational reasoning, scene perception, and modular ADAS control to support the next generation of intelligent driver assistance.

Natalia Matuszczyk, Craig R. Barnes, Rohit Gupta et al.

Bias in geospatial artificial intelligence (GeoAI) models has been documented, yet the evidence is scattered across narrowly focused studies. We synthesize this fragmented literature to provide a concise overview of bias in GeoAI and examine how the EU's Artificial Intelligence Act (EU AI Act) shapes audit obligations. We discuss recurring bias mechanisms, including representation, algorithmic and aggregation bias, and map them to specific provisions of the EU AI Act. By applying the Act's high-risk criteria, we demonstrate that widely deployed GeoAI applications qualify as high-risk systems. We then present examples of recent audits along with an outline of practical methods for detecting bias. As far as we know, this study represents the first integration of GeoAI bias evidence into the EU AI Act context, by identifying high-risk GeoAI systems and mapping bias mechanisms to the Act's Articles. Although the analysis is exploratory, it suggests that even well-curated European datasets should employ routine bias audits before 2027, when the AI Act's high-risk provisions take full effect.

Can Cui, Yunsheng Ma, Sung-Yeon Park et al.

With the broader adoption and highly successful development of Large Language Models (LLMs), there has been growing interest and demand for applying LLMs to autonomous driving technology. Driven by their natural language understanding and reasoning capabilities, LLMs have the potential to enhance various aspects of autonomous driving systems, from perception and scene understanding to interactive decision-making. In this paper, we first introduce the novel concept of designing Large Language Models for Autonomous Driving (LLM4AD), followed by a review of existing LLM4AD studies. Then, we propose a comprehensive benchmark for evaluating the instruction-following and reasoning abilities of LLM4AD systems, which includes LaMPilot-Bench, CARLA Leaderboard 1.0 Benchmark in simulation and NuPlanQA for multi-view visual question answering. Furthermore, we conduct extensive real-world experiments on autonomous vehicle platforms, examining both on-cloud and on-edge LLM deployment for personalized decision-making and motion control. Next, we explore the future trends of integrating language diffusion models into autonomous driving, exemplified by the proposed ViLaD (Vision-Language Diffusion) framework. Finally, we discuss the main challenges of LLM4AD, including latency, deployment, security and privacy, safety, trust and transparency, and personalization.

Ziye Qin, Siyan Li, Guoyuan Wu et al.

Dilemma zones at signalized intersections present a commonly occurring but unsolved challenge for both drivers and traffic operators. Onsets of the yellow lights prompt varied responses from different drivers: some may brake abruptly, compromising the ride comfort, while others may accelerate, increasing the risk of red-light violations and potential safety hazards. Such diversity in drivers' stop-or-go decisions may result from not only surrounding traffic conditions, but also personalized driving behaviors. To this end, identifying personalized driving behaviors and integrating them into advanced driver assistance systems (ADAS) to mitigate the dilemma zone problem presents an intriguing scientific question. In this study, we employ a game engine-based (i.e., CARLA-enabled) driving simulator to collect high-resolution vehicle trajectories, incoming traffic signal phase and timing information, and stop-or-go decisions from four subject drivers in various scenarios. This approach allows us to analyze personalized driving behaviors in dilemma zones and develop a Personalized Transformer Encoder to predict individual drivers' stop-or-go decisions. The results show that the Personalized Transformer Encoder improves the accuracy of predicting driver decision-making in the dilemma zone by 3.7% to 12.6% compared to the Generic Transformer Encoder, and by 16.8% to 21.6% over the binary logistic regression model.

Yunsheng Ma, Wenqian Ye, Xu Cao et al.

Driver intention prediction seeks to anticipate drivers' actions by analyzing their behaviors with respect to surrounding traffic environments. Existing approaches primarily focus on late-fusion techniques, and neglect the importance of maintaining consistency between predictions and prevailing driving contexts. In this paper, we introduce a new framework called Cross-View Episodic Memory Transformer (CEMFormer), which employs spatio-temporal transformers to learn unified memory representations for an improved driver intention prediction. Specifically, we develop a spatial-temporal encoder to integrate information from both in-cabin and external camera views, along with episodic memory representations to continuously fuse historical data. Furthermore, we propose a novel context-consistency loss that incorporates driving context as an auxiliary supervision signal to improve prediction performance. Comprehensive experiments on the Brain4Cars dataset demonstrate that CEMFormer consistently outperforms existing state-of-the-art methods in driver intention prediction.

Ishan Dave, Naman Biyani, Brandon Clark et al.

This technical report presents our approach "Knights" to solve the action recognition task on a small subset of Kinetics-400 i.e. Kinetics400ViPriors without using any extra-data. Our approach has 3 main components: state-of-the-art Temporal Contrastive self-supervised pretraining, video transformer models, and optical flow modality. Along with the use of standard test-time augmentation, our proposed solution achieves 73% on Kinetics400ViPriors test set, which is the best among all of the other entries Visual Inductive Priors for Data-Efficient Computer Vision's Action Recognition Challenge, ICCV 2021.

Rohit Gupta, Sieglinde Kranz, Nikolaus Regnat et al.

Domain-Specific Languages (DSLs) help practitioners in contributing solutions to challenges of specific domains. The efficient development of user-friendly DSLs suitable for industrial practitioners with little expertise in modelling still is challenging. For such practitioners, who often do not model on a daily basis, there is a need to foster reduction of repetitive modelling tasks and providing simplified visual representations of DSL parts. For industrial language engineers, there is no methodical support for providing such guidelines or documentation as part of reusable language modules. Previous research either addresses the reuse of languages or guidelines for modelling. For the efficient industrial deployment of DSLs, their combination is essential: the efficient engineering of DSLs from reusable modules that feature integrated documentation and guidelines for industrial practitioners. To solve these challenges, we propose a systematic approach for the industrial engineering of DSLs based on the concept of reusable DSL Building Blocks, which rests on several years of experience in the industrial engineering of DSLs and their deployment to various organizations. We investigated our approach via focus group methods consisting of five participants from industry and research qualitatively. Ultimately, DSL Building Blocks support industrial language engineers in developing better usable DSLs and industrial practitioners in more efficiently achieving their modelling.

Xiaoyu Zhang, Ajmal Mian, Rohit Gupta et al.

Deep neural networks are being widely deployed for many critical tasks due to their high classification accuracy. In many cases, pre-trained models are sourced from vendors who may have disrupted the training pipeline to insert Trojan behaviors into the models. These malicious behaviors can be triggered at the adversary's will and hence, cause a serious threat to the widespread deployment of deep models. We propose a method to verify if a pre-trained model is Trojaned or benign. Our method captures fingerprints of neural networks in the form of adversarial perturbations learned from the network gradients. Inserting backdoors into a network alters its decision boundaries which are effectively encoded in their adversarial perturbations. We train a two stream network for Trojan detection from its global ($L_\infty$ and $L_2$ bounded) perturbations and the localized region of high energy within each perturbation. The former encodes decision boundaries of the network and latter encodes the unknown trigger shape. We also propose an anomaly detection method to identify the target class in a Trojaned network. Our methods are invariant to the trigger type, trigger size, training data and network architecture. We evaluate our methods on MNIST, NIST-Round0 and NIST-Round1 datasets, with up to 1,000 pre-trained models making this the largest study to date on Trojaned network detection, and achieve over 92\% detection accuracy to set the new state-of-the-art.

Rohit Gupta, Mubarak Shah

Accurate and fine-grained information about the extent of damage to buildings is essential for directing Humanitarian Aid and Disaster Response (HADR) operations in the immediate aftermath of any natural calamity. In recent years, satellite and UAV (drone) imagery has been used for this purpose, sometimes aided by computer vision algorithms. Existing Computer Vision approaches for building damage assessment typically rely on a two stage approach, consisting of building detection using an object detection model, followed by damage assessment through classification of the detected building tiles. These multi-stage methods are not end-to-end trainable, and suffer from poor overall results. We propose RescueNet, a unified model that can simultaneously segment buildings and assess the damage levels to individual buildings and can be trained end-toend. In order to to model the composite nature of this problem, we propose a novel localization aware loss function, which consists of a Binary Cross Entropy loss for building segmentation, and a foreground only selective Categorical Cross-Entropy loss for damage classification, and show significant improvement over the widely used Cross-Entropy loss. RescueNet is tested on the large scale and diverse xBD dataset and achieves significantly better building segmentation and damage classification performance than previous methods and achieves generalization across varied geographical regions and disaster types.

Rohit Gupta, Rohit Panda

Advertisements generate huge chunks of revenues for websites and online businesses. Ad-blocker and tracker blocking programs have gained momentum in the last few years with massive debates raging on privacy concerns and improving user experience online. Acceptable Ads programme and Anti Ad-blockers are primary elements emerging in recent years that combat ad-blockers. In this paper, we discuss at length data collection of top websites in the world, Germany, DACH region and news category. We generate feature based A/B testing metrics and employ classifier evaluations on them along with then analysing the result. Our paper also discusses how Anti Ad-blockers impact the economic, legal and ethical usage in Germany along with the recent changes in GDPR while taking a look at Acceptable ads programme and Whitelisting.

Rohit Gupta, Laurent Besacier, Marc Dymetman et al.

Character-based translation has several appealing advantages, but its performance is in general worse than a carefully tuned BPE baseline. In this paper we study the impact of character-based input and output with the Transformer architecture. In particular, our experiments on EN-DE show that character-based Transformer models are more robust than their BPE counterpart, both when translating noisy text, and when translating text from a different domain. To obtain comparable BLEU scores in clean, in-domain data and close the gap with BPE-based models we use known techniques to train deeper Transformer models.

Rohit Gupta, Patrik Lambert, Raj Nath Patel et al.

As a commercial provider of machine translation, we are constantly training engines for a variety of uses, languages, and content types. In each case, there can be many variables, such as the amount of training data available, and the quality requirements of the end user. These variables can have an impact on the robustness of Neural MT engines. On the whole, Neural MT cures many ills of other MT paradigms, but at the same time, it has introduced a new set of challenges to address. In this paper, we describe some of the specific issues with practical NMT and the approaches we take to improve model robustness in real-world scenarios.

Uroš Kalabić, Rohit Gupta, Stefano Di Cairano et al.

We develop a model predictive control (MPC) design for systems with discrete-time dynamics evolving on smooth manifolds. We show that the properties of conventional MPC for dynamics evolving on $\mathbb R^n$ are preserved and we develop a design procedure for achieving similar properties. We also demonstrate that for discrete-time dynamics on manifolds with Euler characteristic not equal to 1, there do not exist globally stabilizing, continuous control laws. The MPC law is able to achieve global asymptotic stability on these manifolds, because the MPC law may be discontinuous. We apply the method to spacecraft attitude control, where the spacecraft attitude evolves on the Lie group SO(3) and for which a continuous globally stabilizing control law does not exist. In this case, the MPC law is discontinuous and achieves global stability.