Özet:

Object detection is the most common application of computer vision for the last 20 years. It has been widely used for quick & accurate identification and locating of a large number of objects from predefined image categories, real-time video frames, etc. The computer vision field requires several algorithms for detecting objects such as Single-shot detection (SSD), Faster region-based convolutional neural networks (faster R-CNN) & You Only Look Once (YOLO) with its variations using deep learning, etc. Based on parameters such as accuracy, precision & F-score performance of these algorithms is analyzed. In this paper mentioned comments are based on studied literature & key issues related to the topic are also identified which are relevant to the object detection area with accuracy and performance. The paper review begins with an introduction to deep learning and its techniques. (Allamki & Sateesha,2019) Further, CNN architecture and object detection are described along with some modifications to improve performance. Image classification is come into existence for decreasing the gap between human vision and computer vision by training the computer with data. At the end of the article; some promising guidelines & directions are provided for further work in deep learning & object detection techniques.

Anahtar Kelimeler: