Integrating natural language understanding with image perception has led to the development of large vision language models (LVLMs), which showcase remarkable reasoning capabilities. Despite their progress, LVLMs often encounter challenges in accurately anchoring generated text to visual inputs, manifesting as inaccuracies like hallucinations of non-existent scene elements or misinterpretations of object attributes and relationships. Researchers…

The landscape of image segmentation has been profoundly transformed by the introduction of the Segment Anything Model (SAM), a paradigm known for its remarkable zero-shot segmentation capability. SAM’s deployment across a wide array of applications, from augmented reality to data annotation, underscores its utility. However, SAM’s computational intensity, particularly its image encoder’s demand of 2973…

In artificial intelligence, integrating multimodal inputs for video reasoning stands as a frontier, challenging yet ripe with potential. Researchers increasingly focus on leveraging diverse data types – from visual frames and audio snippets to more complex 3D point clouds – to enrich AI’s understanding and interpretation of the world. This endeavor aims to mimic human…

The progress and development of artificial intelligence (AI) heavily rely on human evaluation, guidance, and expertise. In computer vision, convolutional networks acquire a semantic understanding of images through extensive labeling provided by experts, such as delineating object boundaries in datasets like COCO or categorizing images in ImageNet.  Similarly, in robotics, reinforcement learning often relies on…

Artificial intelligence has significantly advanced in developing systems that can interpret and respond to multimodal data. At the forefront of this innovation is Lumos, a groundbreaking multimodal question-answering system designed by researchers at Meta Reality Labs. Unlike traditional systems, Lumos distinguishes itself by its exceptional ability to extract and understand text from images, enhancing the…

The emergence of Multimodality Large Language Models (MLLMs), such as GPT-4 and Gemini, has sparked significant interest in combining language understanding with various modalities like vision. This fusion offers potential for diverse applications, from embodied intelligence to GUI agents. Despite the rapid development of open-source MLLMs like BLIP and LLaMA-Adapter, their performance could be improved…

The capacity of infographics to strategically arrange and use visual signals to clarify complicated concepts has made them essential for efficient communication. Infographics include various visual elements such as charts, diagrams, illustrations, maps, tables, and document layouts. This has been a long-standing technique that makes the material easier to understand. User interfaces (UIs) on desktop…

Big Vision Language Models (VLMs) trained to comprehend vision have shown viability in broad scenarios like visual question answering, visual grounding, and optical character recognition, capitalizing on the strength of Large Language Models (LLMs) in general knowledge of the world. Humans mark or process the provided photos for convenience and rigor to address the intricate…

In recent years, LMMs have rapidly expanded, leveraging CLIP as a foundational vision encoder for robust visual representations and LLMs as versatile tools for reasoning across various modalities. However, while LLMs have grown to over 100 billion parameters, the vision models they rely on need to be bigger, hindering their potential. Scaling up contrastive language-image…

Personalized image generation is the process of generating images of certain personal objects in different user-specified contexts. For example, one may want to visualize the different ways their pet dog would look in different scenarios. Apart from personal experiences, this method also has use cases in personalized storytelling, interactive designs, etc. Although current text-to-image generation…