Learning from the Tangram to Solve Mini Visual Tasks

TL;DR

This paper introduces the Tangram dataset and demonstrates that pre-training on abstract Tangram puzzles enhances neural network performance on various mini visual tasks involving low-resolution images, such as aesthetic assessments and icon recognition.

Contribution

The work creates a novel Tangram dataset inspired by human puzzle-solving, showing that pre-training on it improves performance on diverse low-resolution visual tasks.

Findings

Pre-training on Tangram aids in solving aesthetic tasks like folding clothes and room layout evaluation.

It accelerates few-shot learning in human handwriting recognition.

It improves icon identification accuracy based on contours.

Abstract

Current pre-training methods in computer vision focus on natural images in the daily-life context. However, abstract diagrams such as icons and symbols are common and important in the real world. This work is inspired by Tangram, a game that requires replicating an abstract pattern from seven dissected shapes. By recording human experience in solving tangram puzzles, we present the Tangram dataset and show that a pre-trained neural model on the Tangram helps solve some mini visual tasks based on low-resolution vision. Extensive experiments demonstrate that our proposed method generates intelligent solutions for aesthetic tasks such as folding clothes and evaluating room layouts. The pre-trained feature extractor can facilitate the convergence of few-shot learning tasks on human handwriting and improve the accuracy in identifying icons by their contours. The Tangram dataset is available at https://github.com/yizhouzhao/Tangram.