Object detection, whose main task is to detect objects in a picture to determine the type, location, and scene to which they belong, has become one of the most central problems in computer vision. Object detection is currently broadly divided into two implementations, one-stage detector will get the prediction results directly from the image, two-stage detector is also known as Region-based, because they will regionally generate the image and then submit it to the convolutional neural network. These two approaches have their own advantages and disadvantages, and this paper will study their characteristics and summarize them with examples. This paper summarizes a total of five outstanding works on these two approaches, including YOLO, ConerNet, R-CNN, SPPNet, and Auto-FPN. The summary of these five efforts may help others better understand the field.
Object Detection Algorithms: A Comparison
2022-10-12
1314956 byte
Conference paper
Electronic Resource
English
Object Detection Algorithms - A Review
IEEE | 2023
|Comparison on Detection Algorithms of Small Object Intrusion on High-Speed Railway
Springer Verlag | 2022
|Comparison on Detection Algorithms of Small Object Intrusion on High-Speed Railway
British Library Conference Proceedings | 2022
|