Abstract
Social networks such as Facebook and Instagram are now full of pictures of people’s food and drink images. But it’s not easy to create a successful photography when we take a picture of a great-looking meal. We often shoot unappealing images though the food may look amazing. Food photography is an art not an exact science. There are many food photography apps providing specific functions with a great range of editing features you can adjust anything from exposure to saturation on specific points of the image. In this study, we plan to discuss the tips of guidelines that will help us get better pictures of food. We also want to use deep learning algorithms to learn to judge whether a food image is shot good or not. More than 300 food images captured from social networks are collected as dataset. And each image is marked as one of three different levels: good, fair and poor by experts. A convolutional neural network (CNN) is applied to assess image quality of these food images after manual training. The future application of this study is to design an APPs which can assess the food image quality when users take the camera of mobile phone to shoot a meal.
Details
Presentation Type
Theme
2018 Special Focus: Digital Food Cultures
KEYWORDS
Food Photography, Deep Learning
Digital Media
This presenter hasn’t added media.
Request media and follow this presentation.