PerSense Delivers Expert-Level Instance Recognition Without Any Training

The PerSense framework, a model-agnostic, one-shot, training-free method for customized instance segmentation in dense pictures, is described in this portion of the article. A Few-Shot Object Counter (FSOC) is used to generate density maps, an Instance Detection Module (IDM) is used to identify potential prompts, a Point Prompt Selection Module (PPSM) is used for adaptive filtering, and a Vision-Language Model (VLM) is used for semantic class-label extraction.