Large Vision-Language Models (LVLMs), due to the remarkable visual reasoning ability to understand images and videos, have received widespread attention in the autonomous driving domain, which significantly advances the development of interpretable end-to-end autonomous driving. However, current evaluations of LVLMs primarily focus on the multi-faceted capabilities in common scenarios, lacking quantifiable and automated assessment in autonomous driving contexts, let alone severe road corner cases that even the state-of-the-art autonomous driving perception systems struggle to handle. In this paper, we propose CODA-LM, a novel vision-language benchmark for self-driving, which provides the first automatic and quantitative evaluation of LVLMs for interpretable autonomous driving including general perception, regional perception, and driving suggestions. CODA-LM utilizes the texts to describe the road images, exploiting powerful text-only large language models (LLMs) without image inputs to assess the capabilities of LVLMs in autonomous driving scenarios, which reveals stronger alignment with human preferences than LVLM judges. Experiments demonstrate that even the closed-sourced commercial LVLMs like GPT-4V cannot deal with road corner cases well, suggesting that we are still far from a strong LVLM-powered intelligent driving agent, and we hope our CODA-LM can become the catalyst to promote future development.
Neural Radiance Fields (NeRF) have shown impressive capabilities for photorealistic novel view synthesis when trained on dense inputs. However, when trained on sparse inputs, NeRF typically encounters issues of incorrect density or color predictions, mainly due to insufficient coverage of the scene causing partial and sparse supervision, thus leading to significant performance degradation. While existing works mainly consider ray-level consistency to construct 2D learning regularization based on rendered color, depth, or semantics on image planes, in this paper we propose a novel approach that models 3D spatial field consistency to improve NeRF's performance with sparse inputs. Specifically, we first adopt a voxel-based ray sampling strategy to ensure that the sampled rays intersect with a certain voxel in 3D space. We then randomly sample additional points within the voxel and apply a Transformer to infer the properties of other points on each ray, which are then incorporated into the volume rendering. By backpropagating through the rendering loss, we enhance the consistency among neighboring points. Additionally, we propose to use a contrastive loss on the encoder output of the Transformer to further improve consistency within each voxel. Experiments demonstrate that our method yields significant improvement over different radiance fields in the sparse inputs setting, and achieves comparable performance with current works.
Multimodal large language models (MLLMs) have shown impressive reasoning abilities, which, however, are also more vulnerable to jailbreak attacks than their LLM predecessors. Although still capable of detecting unsafe responses, we observe that safety mechanisms of the pre-aligned LLMs in MLLMs can be easily bypassed due to the introduction of image features. To construct robust MLLMs, we propose ECSO(Eyes Closed, Safety On), a novel training-free protecting approach that exploits the inherent safety awareness of MLLMs, and generates safer responses via adaptively transforming unsafe images into texts to activate intrinsic safety mechanism of pre-aligned LLMs in MLLMs. Experiments on five state-of-the-art (SoTA) MLLMs demonstrate that our ECSO enhances model safety significantly (e.g., a 37.6% improvement on the MM-SafetyBench (SD+OCR), and 71.3% on VLSafe for the LLaVA-1.5-7B), while consistently maintaining utility results on common MLLM benchmarks. Furthermore, we show that ECSO can be used as a data engine to generate supervised-finetuning (SFT) data for MLLM alignment without extra human intervention.
Current perceptive models heavily depend on resource-intensive datasets, prompting the need for innovative solutions. Leveraging recent advances in diffusion models, synthetic data, by constructing image inputs from various annotations, proves beneficial for downstream tasks. While prior methods have separately addressed generative and perceptive models, DetDiffusion, for the first time, harmonizes both, tackling the challenges in generating effective data for perceptive models. To enhance image generation with perceptive models, we introduce perception-aware loss (P.A. loss) through segmentation, improving both quality and controllability. To boost the performance of specific perceptive models, our method customizes data augmentation by extracting and utilizing perception-aware attribute (P.A. Attr) during generation. Experimental results from the object detection task highlight DetDiffusion's superior performance, establishing a new state-of-the-art in layout-guided generation. Furthermore, image syntheses from DetDiffusion can effectively augment training data, significantly enhancing downstream detection performance.
Masked Autoencoder~(MAE) is a prevailing self-supervised learning method that achieves promising results in model pre-training. However, when the various downstream tasks have data distributions different from the pre-training data, the semantically irrelevant pre-training information might result in negative transfer, impeding MAE's scalability. To address this issue, we propose a novel MAE-based pre-training paradigm, Mixture of Cluster-conditional Experts (MoCE), which can be trained once but provides customized pre-training models for diverse downstream tasks. Different from the mixture of experts (MoE), our MoCE trains each expert only with semantically relevant images by using cluster-conditional gates. Thus, each downstream task can be allocated to its customized model pre-trained with data most similar to the downstream data. Experiments on a collection of 11 downstream tasks show that MoCE outperforms the vanilla MAE by 2.45\% on average. It also obtains new state-of-the-art self-supervised learning results on detection and segmentation.
In this paper, we focus on a realistic yet challenging task, Single Domain Generalization Object Detection (S-DGOD), where only one source domain's data can be used for training object detectors, but have to generalize multiple distinct target domains. In S-DGOD, both high-capacity fitting and generalization abilities are needed due to the task's complexity. Differentiable Neural Architecture Search (NAS) is known for its high capacity for complex data fitting and we propose to leverage Differentiable NAS to solve S-DGOD. However, it may confront severe over-fitting issues due to the feature imbalance phenomenon, where parameters optimized by gradient descent are biased to learn from the easy-to-learn features, which are usually non-causal and spuriously correlated to ground truth labels, such as the features of background in object detection data. Consequently, this leads to serious performance degradation, especially in generalizing to unseen target domains with huge domain gaps between the source domain and target domains. To address this issue, we propose the Generalizable loss (G-loss), which is an OoD-aware objective, preventing NAS from over-fitting by using gradient descent to optimize parameters not only on a subset of easy-to-learn features but also the remaining predictive features for generalization, and the overall framework is named G-NAS. Experimental results on the S-DGOD urban-scene datasets demonstrate that the proposed G-NAS achieves SOTA performance compared to baseline methods. Codes are available at https://github.com/wufan-cse/G-NAS.
Although significant progress has been made in the field of 2D-based interactive editing, fine-grained 3D-based interactive editing remains relatively unexplored. This limitation can be attributed to two main challenges: the lack of an efficient 3D representation robust to different modifications and the absence of an effective 3D interactive segmentation method. In this paper, we introduce a novel fine-grained interactive 3D segmentation and editing algorithm with radiance fields, which we refer to as SERF. Our method entails creating a neural mesh representation by integrating multi-view algorithms with pre-trained 2D models. Building upon this representation, we introduce a novel surface rendering technique that preserves local information and is robust to deformation. Moreover, this representation forms the basis for achieving accurate and interactive 3D segmentation without requiring 3D supervision. Harnessing this representation facilitates a range of interactive 3D editing operations, encompassing tasks such as interactive geometry editing and texture painting. Extensive experiments and visualization examples of editing on both real and synthetic data demonstrate the superiority of our method on representation quality and editing ability.
Instruction tuning of the Large Vision-language Models (LVLMs) has revolutionized the development of versatile models with zero-shot generalization across a wide range of downstream vision-language tasks. However, diversity of training tasks of different sources and formats would lead to inevitable task conflicts, where different tasks conflicts for the same set of model parameters, resulting in sub-optimal instruction-following abilities. To address that, we propose the Mixture of Cluster-conditional LoRA Experts (MoCLE), a novel Mixture of Experts (MoE) architecture designed to activate the task-customized model parameters based on the instruction clusters. A separate universal expert is further incorporated to improve the generalization capabilities of MoCLE for novel instructions. Extensive experiments on 10 zero-shot tasks demonstrate the effectiveness of MoCLE.
Large language models (LLMs) have shown remarkable proficiency in human-level reasoning and generation capabilities, which encourages extensive research on their application in mathematical problem solving. However, current work has been largely focused on text-based mathematical problems, with limited investigation in problems involving geometric information. Addressing this gap, we aim to enable LLMs to solve geometric problems by understanding image input. We first analyze the limitations of current Multimodal Large Language Models (MLLMs) in this area: they struggle to accurately comprehending basic geometric elements and their relationships. To overcome these challenges, we take advantage of the unique characteristics of geometric problems (such as unique geometric logical form, and geometric scalability) and the capacity of the textual LLMs to build an enriched multimodal geometry dataset based on existing data. The augmented dataset, Geo170K, contains more than 170K geometric image-caption and question-answer pairs. Utilizing our constructed Geo170K dataset, we develop G-LLaVA, which demonstrates exceptional performance in solving geometric problems, significantly outperforming GPT-4-V on the MathVista benchmark with only 7B parameters.