Automatic dataset generation for specific object detection

TL;DR

This paper introduces a novel method for automatically generating object detection datasets by synthesizing images with objects seamlessly integrated into new scenes, reducing manual labeling effort and enhancing data quality.

Contribution

It presents a new approach to synthesize object-in-scene images that preserve detailed features and improve training data for detection models, addressing scalability and label limitations.

Findings

Synthesized images blend object boundaries effectively with backgrounds.

State-of-the-art segmentation models perform well on the generated data.

The method reduces the need for manual dataset creation.

Abstract

In the past decade, object detection tasks are defined mostly by large public datasets. However, building object detection datasets is not scalable due to inefficient image collecting and labeling. Furthermore, most labels are still in the form of bounding boxes, which provide much less information than the real human visual system. In this paper, we present a method to synthesize object-in-scene images, which can preserve the objects' detailed features without bringing irrelevant information. In brief, given a set of images containing a target object, our algorithm first trains a model to find an approximate center of the object as an anchor, then makes an outline regression to estimate its boundary, and finally blends the object into a new scene. Our result shows that in the synthesized image, the boundaries of objects blend very well with the background. Experiments also show that SOTA segmentation models work well with our synthesized data.