Plain-Det: A Plain Multi-Dataset Object Detector

TL;DR

Plain-Det is a versatile multi-dataset object detection framework that enhances training efficiency and robustness, achieving state-of-the-art results on COCO and demonstrating strong generalization across 13 datasets.

Contribution

It introduces a flexible, robust, and efficient multi-dataset object detector compatible with various architectures, improving performance and generalization in dense vision tasks.

Findings

Achieves 51.9 mAP on COCO using Def-DETR with Plain-Det

Demonstrates strong generalization across 13 datasets

Compatible with various detection architectures

Abstract

Recent advancements in large-scale foundational models have sparked widespread interest in training highly proficient large vision models. A common consensus revolves around the necessity of aggregating extensive, high-quality annotated data. However, given the inherent challenges in annotating dense tasks in computer vision, such as object detection and segmentation, a practical strategy is to combine and leverage all available data for training purposes. In this work, we propose Plain-Det, which offers flexibility to accommodate new datasets, robustness in performance across diverse datasets, training efficiency, and compatibility with various detection architectures. We utilize Def-DETR, with the assistance of Plain-Det, to achieve a mAP of 51.9 on COCO, matching the current state-of-the-art detectors. We conduct extensive experiments on 13 downstream datasets and Plain-Det demonstrates strong generalization capability. Code is release at https://github.com/ChengShiest/Plain-Det