Deep Learning Photo Aesthetics
Tags: Computer Vision Deep Learning TensorFlow AestheticsAll of my successes, and failures, on the path to developing a deep learning model capable of understanding photo aesthetics.
- Introduction and Data Analysis - 26 October 2020
- Data Preprocessing - 19 November 2020
- Data Pipeline Optimization - 17 December 2020
My analysis of the AVA database as the first step in creating a deep learning model capable understanding photo aesthetics.
How I shaped the AVA database to reduce training times, remove bias, and improve model inference.
Optimizing image-based TFRecords to reduce storage space and improve training times.