Research for Object Detection Based on Deep Learning

Object detection, which is one of the most essential and challenging issue in the computer vision, aims to classify the instances of objects in the picture to the right classes and locate them correctly from natural image. In this paper, we will focus on two state-of-art object detection algorithms (Single Shot MultiBox Detector and Cascaded CNNs & Adversarial Learning), which achieve great improvement recent years and implement them to practical dataset to see their performance. Generally, there are some popular application programming interface (API) in the implement of object detection algorithms, such as PyTorch, Keras, and TenserFlow.