In this article, we explore Fast Segment Anything, a CNN based alternative approach to the original Transformer based Segment Anything Model. ...

In this article, we explore HQ-SAM, a modified version of SAM that overcomes some of the limitations when trying to segment small and intricate objects. ...

In this article, we create a custom Vision Transformer based object detection model using NVIDIA's FasterViT backbone and the Single Shot Detection head. ...

In this article, we integrate SAM2, Molmo, and Whisper for creating a text-based as well as speech-to-text pipeline for automated object segmentation in images. ...

In this article, we use SAM2 and Molmo for carrying out image segmentation using natural language. We provide a prompt to Molmo, get the coordinates, and pass these to SAM2.1 to segment the objects ...