Survey on Object Detection Techniques

Shrikant A. Shinde¹, Akshay Mali², Aniket Thorat³, Sourabh Khot⁴

Department of Computer Engineering

Sinhgad Institute of Technology and Science, Narhe Pune, Maharashtra, India

Abstract: Image classification is assigning a class label to an image on the other hand object localization requires drawing a bounding box around one or more objects in an image. Object detection is more difficult and combines these two tasks and draws a bounding box around each object of interest in the image and assigns them a class label. All this is referred to as object recognition. The proposed smart cart allows the customer to shop without any friction. A product that is picked and placed in the cart by the customer is recognized by the cart with help of sensors. Using yolo for product recognition and other sensors the data is fused. The type, quantity and weight is automatically calculated and displayed on the screen present above the cart. Customer can also remove an item from the cart which will be immediately reflected in the virtual cart. Based on the customers shopping history recommendations can be suggested also an instore map is provided. Barcode’s don't need to be scanned or searched. By simply picking an item and moving on allows the customer to shop quickly and easily while not worrying about waiting in the checkout lines. After the shopping is complete the bill is generated and paid by the customer using online methods.

Keywords: Yolo, Object Detection