Automatic image cropping with Faster_RCNN

• Main Author: Vu, Ha Son
• Other Authors: Chia Liang Tien
• Format: Final Year Project
• Language: English
• Published: 2017
• Subjects: DRNTU::Engineering::Computer science and engineering::Computing methodologies::Artificial intelligence; DRNTU::Engineering::Computer science and engineering::Computing methodologies::Image processing and computer vision
• Online Access: http://hdl.handle.net/10356/72799
• Institution: Nanyang Technological University

Convolutional Neural Networks have been proven useful in many computer vision tasks such as image classification and object detection. On the other hand, automatic image cropping remains a challenging task given its subjective nature. This project explored the performance of an object detection method, Faster R-CNN, in doing automatic image cropping task to enhance image composition. The focus of the study is on three common compositional rules: Leading Lines, Space-to-move and Symmetry/Reflection. The final model was subsequently used to build a web application that helped inexperienced photographers to do cropping to enhance their image composition according to the three chosen rules.