User Constrained Thumbnail Generation using Adaptive Convolutions
This addresses the need for flexible thumbnail generation in digital media, though it appears incremental as it builds on existing methods like RPN and GCA.
The paper tackles the problem of generating thumbnails of any size and aspect ratio, even unseen during training, by proposing a deep neural framework using adaptive convolutions, achieving high accuracy and precision with superior performance over state-of-the-art techniques.
Thumbnails are widely used all over the world as a preview for digital images. In this work we propose a deep neural framework to generate thumbnails of any size and aspect ratio, even for unseen values during training, with high accuracy and precision. We use Global Context Aggregation (GCA) and a modified Region Proposal Network (RPN) with adaptive convolutions to generate thumbnails in real time. GCA is used to selectively attend and aggregate the global context information from the entire image while the RPN is used to predict candidate bounding boxes for the thumbnail image. Adaptive convolution eliminates the problem of generating thumbnails of various aspect ratios by using filter weights dynamically generated from the aspect ratio information. The experimental results indicate the superior performance of the proposed model over existing state-of-the-art techniques.