In the era of Generative AI and Language Models we are forgetting about the basic building blocks of machine learning. The tools for Text based image generation like DALLE-2, Midjourney, etc. and Chatbots based on Large Language Models (LLMs) like ChatGPT and LLaMA have taken the technology world by storm. Everyone is talking about these tools and there is a hot discussion that AI will replace a lot of mundane and boring jobs. In the midst of thsese developments we have forgotten about basic ML tasks like Image Classification, Image Segmentation, etc.
Meta the parent company of Social Media giant Facebook has launched a Image Segmentation model and Dataset. This model was launched on 5th April 2023. The model is called Segment Anything Model and the largest ever segmentation dataset is called SA-1B Dataset.
The capabalities of SAM (Segment Anything Model) are -
(1) SAM allows users to segment objects with just a click or by interactively clicking points to include and exclude from the object. The model can also be prompted with a bounding box.
(2) SAM can output multiple valid masks when faced with ambiguity about the object being segmented, an important and necessary capability for solving segmentation in the real world.
(3) SAM can automatically find and mask all objects in an image.
(4) SAM can generate a segmentation mask for any prompt in real time after precomputing the image embedding, allowing for real-time interaction with the model.
The SA-1B Dataset includes more than 1.1 billion segmentation masks collected on about 11 million licensed and privacy-preserving images. SA-1B has 400x more masks than any existing segmentation dataset, and as verified by human evaluation studies, the masks are of high quality and diversity.
You can read about in detail here - https://ai.facebook.com/blog/segment-anything-foundation-model-image-segmentation/