BigDetection: A Large-scale Benchmark for Improved Object Detector Pre-training

TL;DR

BigDetection is a new large-scale benchmark dataset with 600 categories and over 3.4 million images, designed to improve object detector pre-training and evaluation.

Contribution

The paper introduces BigDetection, a unified, large-scale dataset combining multiple sources to enhance object detection pre-training and benchmarking.

Findings

BigDetection outperforms previous datasets in pre-training effectiveness.

It provides a comprehensive benchmark for evaluating object detection methods.

The dataset enables training more general and powerful detectors.

Abstract

Multiple datasets and open challenges for object detection have been introduced in recent years. To build more general and powerful object detection systems, in this paper, we construct a new large-scale benchmark termed BigDetection. Our goal is to simply leverage the training data from existing datasets (LVIS, OpenImages and Object365) with carefully designed principles, and curate a larger dataset for improved detector pre-training. Specifically, we generate a new taxonomy which unifies the heterogeneous label spaces from different sources. Our BigDetection dataset has 600 object categories and contains over 3.4M training images with 36M bounding boxes. It is much larger in multiple dimensions than previous benchmarks, which offers both opportunities and challenges. Extensive experiments demonstrate its validity as a new benchmark for evaluating different object detection methods, and its effectiveness as a pre-training dataset.