Intelligent Auto Annotation System for Identifying Vehicles without Number Plates Using Grounding DINO and Segment Anything Model

It is difficult to automatically annotate things in pictures and movies, especially when the items have different properties, such as cars without license plates. In this study, we offer a unique method that combines the Segments Anything Model (SAM) with Grounding DINO (the identification of noisy objects) for the auto-annotation of automobiles without number plates. Modern object identification models like Grounding DINO can detect things with accuracy even in congested or noisy situations. Conversely, SAM is a flexible segmentation model that does not require predetermined classes and can separate objects according to semantic cues. Our method operates by first identifying automobiles in a picture or video frame using grounding DINO. When a car without a license plate is found, SAM is used to segment the car and produce a mask image. For the algorithm to reliably identify and segment such vehicles, it just has to prompt for the thing's name (“vehicle without license plate”). Annotated mask pictures that are produced can be utilized in traffic management or surveillance systems, among other applications, for additional processing or analysis. We test our method on a set of surveillance videos and photos that include cars without license plates. Based on experimental data, we find that our method outperforms existing methods in recognizing and segmenting automobiles without number plates with high accuracy. The proposed auto-annotation system has the potential to enhance the effectiveness and efficiency of traffic control and surveillance systems, particularly in identifying vehicles that are important to law enforcement.