Learning from THEODORE: A Synthetic Omnidirectional Top-View Indoor Dataset for Deep Transfer Learning

TL;DR

This paper introduces THEODORE, a large-scale synthetic omnidirectional indoor dataset with diverse fisheye images, designed to improve deep transfer learning for object detection in indoor environments.

Contribution

The paper presents a novel synthetic dataset with 100,000 high-resolution fisheye images, annotations, and demonstrates its effectiveness for fine-tuning CNNs for object detection.

Findings

High AP of 0.84 for person detection on HD datasets

Synthetic data improves CNN transfer learning for indoor object detection

Domain randomization enhances model generalization

Abstract

Recent work about synthetic indoor datasets from perspective views has shown significant improvements of object detection results with Convolutional Neural Networks(CNNs). In this paper, we introduce THEODORE: a novel, large-scale indoor dataset containing 100,000 high-resolution diversified fisheye images with 14 classes. To this end, we create 3D virtual environments of living rooms, different human characters and interior textures. Beside capturing fisheye images from virtual environments we create annotations for semantic segmentation, instance masks and bounding boxes for object detection tasks. We compare our synthetic dataset to state of the art real-world datasets for omnidirectional images. Based on MS COCO weights, we show that our dataset is well suited for fine-tuning CNNs for object detection. Through a high generalization of our models by means of image synthesis and domain randomization, we reach an AP up to 0.84 for class person on High-Definition Analytics dataset.