Abstract:

In the area of computer vision, real-time object detection is a complex and dynamic task that includes localizing a single object or detecting multiple objects in an image. Modern deep learning models uses different networks for various tasks on embedded systems. In this project, a method has been developed for object recognition using the pre-trained deep learning model [1]. The objective is, to detect objects by capturing images from a webcam and subsequently detecting objects in a video stream which is specifically designed to be efficient for embedded systems [2]. To achieve the objective, the algorithm captures the images from the webcam and processes them in real-time using the Mobile Net model. The model processes the images and detects the objects present in them. This process is repeated for every frame of the video stream, enabling real-time object detection for the entire video. Overall, the developed method for real-time object detection using the Mobile Net model offers a promising solution for efficient and accurate object detection in various applications.

Description:

Neural Networks, OpenCV, Object Detection, Mobile Net, SSD, Machine Learning