Towards Reflected Object Detection: A Benchmark

Zhongtian Wang,You Wu,Hui Zhou,Shuiwang Li

2024-07-08

Abstract:Object detection has greatly improved over the past decade thanks to advances in deep learning and large-scale datasets. However, detecting objects reflected in surfaces remains an underexplored area. Reflective surfaces are ubiquitous in daily life, appearing in homes, offices, public spaces, and natural environments. Accurate detection and interpretation of reflected objects are essential for various applications. This paper addresses this gap by introducing a extensive benchmark specifically designed for Reflected Object Detection. Our Reflected Object Detection Dataset (RODD) features a diverse collection of images showcasing reflected objects in various contexts, providing standard annotations for both real and reflected objects. This distinguishes it from traditional object detection benchmarks. RODD encompasses 10 categories and includes 21,059 images of real and reflected objects across different backgrounds, complete with standard bounding box annotations and the classification of objects as real or reflected. Additionally, we present baseline results by adapting five state-of-the-art object detection models to address this challenging task. Experimental results underscore the limitations of existing methods when applied to reflected object detection, highlighting the need for specialized approaches. By releasing RODD, we aim to support and advance future research on detecting reflected objects. Dataset and code are available at: https: //github.com/Tqybu-hans/RODD.

Computer Vision and Pattern Recognition

What problem does this paper attempt to address?

### Problems Addressed by the Paper

This paper aims to address the problem of detecting objects in reflective surfaces (such as mirrors, glass windows, etc.). Although significant progress has been made in object detection technology based on deep learning in recent years, detecting objects in reflective surfaces remains an under-explored area. To fill this gap, the authors introduce a benchmark dataset specifically for reflected object detection—RODD (Reflected Object Detection Dataset). This dataset includes images of reflected objects in various scenarios and provides standard annotation information, including the distinction between real objects and reflected objects.

Specifically, the RODD dataset includes the following features:
- **Diversity**: Covers 10 categories, with a total of 21,059 images.
- **Detailed Annotations**: Each image comes with axis-aligned bounding boxes, category labels, and object attributes (real or reflected).
- **Benchmark Models**: The paper also provides baseline results based on five state-of-the-art object detection models (such as YOLOv8, YOLOv10, etc.) to evaluate their performance on the task of reflected object detection.

By introducing this benchmark dataset, the authors hope to promote research and development in the field of reflected object detection, improving the performance and reliability of various application scenarios (such as surveillance systems, autonomous driving, service robots, etc.).