Data Augmentation and Faster RCNN Improve Vehicle Detection and Recognition
Reddy Alexandro Harianto, Yuliana Melita Pranoto, Tjwanda Putera Gunawan · 2021
Vehicle detection is a technique that uses photographs to identify and label vehicles. Images for their learning may be taken in various scenarios, such as focus ranges or different lighting. Lower yields may have arisen from the photographs taken under normal conditions. Better accuracy can be achieved by improving the original frames by correcting the point of view conditions. The simulation can test machine learning performance by simulating different scenarios, such as whether the images are good. One method of simulating the initial image conditions is to use data augmentation. Several techniques for machine learning for image recognition have been developed, such as Neural Networks and Deep Learning, which can be used in large image datasets indefinitely. An attribute from each type or class's images is used to classify the type or classes in Deep Learning. Faster R-CNN with Data Augmentation was proposed as a tool for classifying images learned from a large dataset. Data augmentation was used to replicate the images that could be produced to improve precision.