3D self-supervised learning (SSL) has faced persistent challenges in developing semantically meaningful point representations suitable for diverse applications with minimal supervision. Despite substantial progress in image-based SSL, existing point cloud SSL methods have largely been limited due to the issue known as the “geometric shortcut,” where models excessively rely on low-level geometric features like surface…

Recent advancements in AI scaling laws have shifted from merely increasing model size and training data to optimizing inference-time computation. This approach, exemplified by models like OpenAI o1 and DeepSeek R1, enhances model performance by leveraging additional computational resources during inference. Test-time budget forcing has emerged as an efficient technique in LLMs, enabling improved performance…

Process-supervised reward models (PRMs) offer fine-grained, step-wise feedback on model responses, aiding in selecting effective reasoning paths for complex tasks. Unlike output reward models (ORMs), which evaluate responses based on final outputs, PRMs provide detailed assessments at each step, making them particularly valuable for reasoning-intensive applications. While PRMs have been extensively studied in language tasks,…

LLMs have shown impressive capabilities in reasoning tasks like Chain-of-Thought (CoT), enhancing accuracy and interpretability in complex problem-solving. While researchers are extending these capabilities to multi-modal domains, videos present unique challenges due to their temporal dimension. Unlike static images, videos require understanding dynamic interactions over time. Current visual CoT methods excel with static inputs but…

Large Vision-Language Models (LVLMs) have made significant strides in recent years, yet several key limitations persist. One major challenge is aligning these models effectively with human expectations, particularly for tasks involving detailed and precise visual information. Traditionally, LVLMs undergo a two-stage training paradigm: pretraining followed by supervised fine-tuning. However, supervised fine-tuning alone cannot fully overcome…

Stereo depth estimation plays a crucial role in computer vision by allowing machines to infer depth from two images. This capability is vital for autonomous driving, robotics, and augmented reality applications. Despite advancements in deep learning, many existing stereo-matching models require domain-specific fine-tuning to achieve high accuracy. The challenge lies in developing a model that…

Understanding videos with AI requires handling sequences of images efficiently. A major challenge in current video-based AI models is their inability to process videos as a continuous flow, missing important motion details and disrupting continuity. This lack of temporal modeling prevents tracing changes; therefore, events and interactions are partially unknown. Long videos also make the…

With researchers aiming to unify visual generation and understanding into a single framework, multimodal artificial intelligence is evolving rapidly. Traditionally, these two domains have been treated separately due to their distinct requirements. Generative models focus on producing fine-grained image details while understanding models prioritize high-level semantics. The challenge lies in integrating both capabilities effectively without…

Modern vision-language models have transformed how we process visual data, yet they often fall short when it comes to fine-grained localization and dense feature extraction. Many traditional models focus on high-level semantic understanding and zero-shot classification but struggle with detailed spatial reasoning. These limitations can impact applications that require precise localization, such as document analysis…

Multimodal AI agents are designed to process and integrate various data types, such as images, text, and videos, to perform tasks in digital and physical environments. They are used in robotics, virtual assistants, and user interface automation, where they need to understand and act based on complex multimodal inputs. These systems aim to bridge verbal…