Recently, Large Multimodal Models (LMMs) have made significant progress in video question-answering using a frame-wise approach by leveraging large-scale, image-based pretraining in a zero-shot manner. While image-based methods for videos have shown impressive performance, a current limitation is that they often overlook how key timestamps are selected and cannot adjust when incorrect timestamps are identified. Moreover, they are unable to extract details relevant to the question, instead providing general descriptions of the frame. To overcome this, we design a multi-LMM agent framework that travels along the video, iteratively collecting relevant information from keyframes through interactive question-asking until there is sufficient information to answer the question. Specifically, we propose TraveLER, a model that can create a plan to "Traverse" through the video, ask questions about individual frames to "Locate" and store key information, and then "Evaluate" if there is enough information to answer the question. Finally, if there is not enough information, our method is able to "Replan" based on its collected knowledge. Through extensive experiments, we find that the proposed TraveLER approach improves performance on several video question-answering benchmarks, such as NExT-QA, STAR, and Perception Test, without the need to fine-tune on specific datasets.
Robotic access monitoring of multiple target areas has applications including checkpoint enforcement, surveillance and containment of fire and flood hazards. Monitoring access for a single target region has been successfully modeled as a minimum-cut problem. We generalize this model to support multiple target areas using two approaches: iterating on individual targets and examining the collections of targets holistically. Through simulation we measure the performance of each approach on different scenarios.
Multimodal machine learning has gained significant attention in recent years due to its potential for integrating information from multiple modalities to enhance learning and decision-making processes. However, it is commonly observed that unimodal models outperform multimodal models, despite the latter having access to richer information. Additionally, the influence of a single modality often dominates the decision-making process, resulting in suboptimal performance. This research project aims to address these challenges by proposing a novel regularization term that encourages multimodal models to effectively utilize information from all modalities when making decisions. The focus of this project lies in the video-audio domain, although the proposed regularization technique holds promise for broader applications in embodied AI research, where multiple modalities are involved. By leveraging this regularization term, the proposed approach aims to mitigate the issue of unimodal dominance and improve the performance of multimodal machine learning systems. Through extensive experimentation and evaluation, the effectiveness and generalizability of the proposed technique will be assessed. The findings of this research project have the potential to significantly contribute to the advancement of multimodal machine learning and facilitate its application in various domains, including multimedia analysis, human-computer interaction, and embodied AI research.
The segmentation of organs in volumetric medical images plays an important role in computer-aided diagnosis and treatment/surgery planning. Conventional 2D convolutional neural networks (CNNs) can hardly exploit the spatial correlation of volumetric data. Current 3D CNNs have the advantage to extract more powerful volumetric representations but they usually suffer from occupying excessive memory and computation nevertheless. In this study we aim to enhance the 2D networks with contextual information for better volumetric image segmentation. Accordingly, we propose a contextual embedding learning approach to facilitate 2D CNNs capturing spatial information properly. Our approach leverages the learned embedding and the slice-wisely neighboring matching as a soft cue to guide the network. In such a way, the contextual information can be transferred slice-by-slice thus boosting the volumetric representation of the network. Experiments on challenging prostate MRI dataset (PROMISE12) and abdominal CT dataset (CHAOS) show that our contextual embedding learning can effectively leverage the inter-slice context and improve segmentation performance. The proposed approach is a plug-and-play, and memory-efficient solution to enhance the 2D networks for volumetric segmentation. The code will be publicly available.
We formulate active perception for an autonomous agent that explores an unknown environment as a two-player zero-sum game: the agent aims to maximize information gained from the environment while the environment aims to minimize the information gained by the agent. In each episode, the environment reveals a set of actions with their potentially erroneous information gain. In order to select the best action, the robot needs to recover the true information gain from the erroneous one. The robot does so by minimizing the discrepancy between its estimate of information gain and the true information gain it observes after taking the action. We propose an online convex optimization algorithm that achieves sub-linear expected regret $O(T^{3/4})$ for estimating the information gain. We also provide a bound on the regret of active perception performed by any (near-)optimal prediction and trajectory selection algorithms. We evaluate this approach using semantic neural radiance fields (NeRFs) in simulated realistic 3D environments to show that the robot can discover up to 12% more objects using the improved estimate of the information gain. On the M3ED dataset, the proposed algorithm reduced the error of information gain prediction in occupancy map by over 67%. In real-world experiments using occupancy maps on a Jackal ground robot, we show that this approach can calculate complicated trajectories that efficiently explore all occluded regions.
Detecting small moving objects in complex backgrounds from an overhead perspective is a highly challenging task for machine vision systems. As an inspiration from nature, the avian visual system is capable of processing motion information in various complex aerial scenes, and its Retina-OT-Rt visual circuit is highly sensitive to capturing the motion information of small objects from high altitudes. However, more needs to be done on small object motion detection algorithms based on the avian visual system. In this paper, we conducted mathematical modeling based on extensive studies of the biological mechanisms of the Retina-OT-Rt visual circuit. Based on this, we proposed a novel tectum small object motion detection neural network (TSOM). The neural network includes the retina, SGC dendritic, SGC Soma, and Rt layers, each layer corresponding to neurons in the visual pathway. The Retina layer is responsible for accurately projecting input content, the SGC dendritic layer perceives and encodes spatial-temporal information, the SGC Soma layer computes complex motion information and extracts small objects, and the Rt layer integrates and decodes motion information from multiple directions to determine the position of small objects. Extensive experiments on pigeon neurophysiological experiments and image sequence data showed that the TSOM is biologically interpretable and effective in extracting reliable small object motion features from complex high-altitude backgrounds.
Emotional Support Conversation (ESC) systems are pivotal in providing empathetic interactions, aiding users through negative emotional states by understanding and addressing their unique experiences. In this paper, we tackle two key challenges in ESC: enhancing contextually relevant and empathetic response generation through dynamic demonstration retrieval, and advancing cognitive understanding to grasp implicit mental states comprehensively. We introduce Dynamic Demonstration Retrieval and Cognitive-Aspect Situation Understanding (\ourwork), a novel approach that synergizes these elements to improve the quality of support provided in ESCs. By leveraging in-context learning and persona information, we introduce an innovative retrieval mechanism that selects informative and personalized demonstration pairs. We also propose a cognitive understanding module that utilizes four cognitive relationships from the ATOMIC knowledge source to deepen situational awareness of help-seekers' mental states. Our supportive decoder integrates information from diverse knowledge sources, underpinning response generation that is both empathetic and cognitively aware. The effectiveness of \ourwork is demonstrated through extensive automatic and human evaluations, revealing substantial improvements over numerous state-of-the-art models, with up to 13.79\% enhancement in overall performance of ten metrics. Our codes are available for public access to facilitate further research and development.
Recently, we have witnessed impressive achievements in cancer survival analysis by integrating multimodal data, e.g., pathology images and genomic profiles. However, the heterogeneity and high dimensionality of these modalities pose significant challenges for extracting discriminative representations while maintaining good generalization. In this paper, we propose a Cohort-individual Cooperative Learning (CCL) framework to advance cancer survival analysis by collaborating knowledge decomposition and cohort guidance. Specifically, first, we propose a Multimodal Knowledge Decomposition (MKD) module to explicitly decompose multimodal knowledge into four distinct components: redundancy, synergy and uniqueness of the two modalities. Such a comprehensive decomposition can enlighten the models to perceive easily overlooked yet important information, facilitating an effective multimodal fusion. Second, we propose a Cohort Guidance Modeling (CGM) to mitigate the risk of overfitting task-irrelevant information. It can promote a more comprehensive and robust understanding of the underlying multimodal data, while avoiding the pitfalls of overfitting and enhancing the generalization ability of the model. By cooperating the knowledge decomposition and cohort guidance methods, we develop a robust multimodal survival analysis model with enhanced discrimination and generalization abilities. Extensive experimental results on five cancer datasets demonstrate the effectiveness of our model in integrating multimodal data for survival analysis.
Recent developments underscore the potential of textual information in enhancing learning models for a deeper understanding of medical visual semantics. However, language-guided medical image segmentation still faces a challenging issue. Previous works employ implicit and ambiguous architectures to embed textual information. This leads to segmentation results that are inconsistent with the semantics represented by the language, sometimes even diverging significantly. To this end, we propose a novel cross-modal conditioned Reconstruction for Language-guided Medical Image Segmentation (RecLMIS) to explicitly capture cross-modal interactions, which assumes that well-aligned medical visual features and medical notes can effectively reconstruct each other. We introduce conditioned interaction to adaptively predict patches and words of interest. Subsequently, they are utilized as conditioning factors for mutual reconstruction to align with regions described in the medical notes. Extensive experiments demonstrate the superiority of our RecLMIS, surpassing LViT by 3.74% mIoU on the publicly available MosMedData+ dataset and achieving an average increase of 1.89% mIoU for cross-domain tests on our QATA-CoV19 dataset. Simultaneously, we achieve a relative reduction of 20.2% in parameter count and a 55.5% decrease in computational load. The code will be available at https://github.com/ShashankHuang/RecLMIS.
Recent advancements in video saliency prediction (VSP) have shown promising performance compared to the human visual system, whose emulation is the primary goal of VSP. However, current state-of-the-art models employ spatio-temporal transformers trained on limited amounts of data, hindering generalizability adaptation to downstream tasks. The benefits of vision foundation models present a potential solution to improve the VSP process. However, adapting image foundation models to the video domain presents significant challenges in modeling scene dynamics and capturing temporal information. To address these challenges, and as the first initiative to design a VSP model based on video foundation models, we introduce SalFoM, a novel encoder-decoder video transformer architecture. Our model employs UnMasked Teacher (UMT) as feature extractor and presents a heterogeneous decoder which features a locality-aware spatio-temporal transformer and integrates local and global spatio-temporal information from various perspectives to produce the final saliency map. Our qualitative and quantitative experiments on the challenging VSP benchmark datasets of DHF1K, Hollywood-2 and UCF-Sports demonstrate the superiority of our proposed model in comparison with the state-of-the-art methods.