Introduction

Object detection detects one or more objects within an image and assigns each one of them a label, as opposed to image classification which assigns a lable to the whole image. The predicted locations are typically rectangles (aka bounding boxes). If you are looking for shapes or polygons, then have a look at instance segmentation.

What to do next:

• Annotate your data
• Choose a framework