Glass Surface Detection: Leveraging Reflection Dynamics in Flash/No-flash Imagery

TL;DR

Glass surfaces are difficult to detect due to transparency and weak features. The authors introduce NFGlassNet, a no-flash and flash image pair driven detector that uses Reflection Contrast Mining (RCMM) and Reflection Guided Attention (RGAM) to extract and fuse reflection cues with glass features, enabling accurate localization. A new NFGD dataset (~3.3k pairs) supports learning of reflection appearance/disappearance under varied illumination. Experiments show NFGlassNet surpasses state-of-the-art methods on multiple benchmarks, validating the effectiveness of reflection-based cues for glass surface detection.

Abstract

Glass surfaces are ubiquitous in daily life, typically appearing colorless, transparent, and lacking distinctive features. These characteristics make glass surface detection a challenging computer vision task. Existing glass surface detection methods always rely on boundary cues (e.g., window and door frames) or reflection cues to locate glass surfaces, but they fail to fully exploit the intrinsic properties of the glass itself for accurate localization. We observed that in most real-world scenes, the illumination intensity in front of the glass surface differs from that behind it, which results in variations in the reflections visible on the glass surface. Specifically, when standing on the brighter side of the glass and applying a flash towards the darker side, existing reflections on the glass surface tend to disappear. Conversely, while standing on the darker side and applying a flash towards the brighter side, distinct reflections will appear on the glass surface. Based on this phenomenon, we propose NFGlassNet, a novel method for glass surface detection that leverages the reflection dynamics present in flash/no-flash imagery. Specifically, we propose a Reflection Contrast Mining Module (RCMM) for extracting reflections, and a Reflection Guided Attention Module (RGAM) for fusing features from reflection and glass surface for accurate glass surface detection. For learning our network, we also construct a dataset consisting of 3.3K no-flash and flash image pairs captured from various scenes with corresponding ground truth annotations. Extensive experiments demonstrate that our method outperforms the state-of-the-art methods. Our code, model, and dataset will be available upon acceptance of the manuscript.

Figures (17)

• Figure 1: Comparison of our NFGlassNet with the state-of-the-art glass surface detection methods. For each scene from left to right: the first and second columns are no-flash image and flash image, respectively, and the rest columns are results produced by the competing methods and our method. GSDNet lin2021rich tends to a under-detection when there is no reflection on the glass surface ($1st$ scene, no-flash image). Both GSDNet lin2021rich and RGB-T huo2023glass are prone to over-detection when there are numerous glass-like frame ($2nd$ scene), as they heavily rely on the boundary cues. Our method can effectively leverage the appearance and disappearance of reflections in the same scene, which enables more accurate identification of glass regions. The red arrows indicate the locations of reflections, and zooming in on the images can provide a better visual effect.
• Figure 2: Scenes rendering by Blender for appearance ($1st$ scene) and disappearance ($2nd$ scene) of reflection: Distinct reflection on the glass surface will disappear when take photographs from the high-lighted side and apply flash towards the darker side (from Fig. \ref{['fig:light_side_no_flash']} to Fig. \ref{['fig:light_side_flash']}). Conversely, in some situations, clear reflection will appear on the glass surface when we take photographs from the low-lighted side and apply flash towards the light side (from Fig. \ref{['fig:dark_side_no_flash']} to Fig. \ref{['fig:dark_side_flash']}).
• Figure 3: Comparison between GSD lin2021rich, GSGD Yan2025ghosting and our NFGD. Each no-flash and flash image pair from our NFGD exhibits the appearance and disappearance phenomenon of reflections on glass surfaces. Our NFGD encompasses a wide range of shots across various real-world scenes.
• Figure 4: Statistics of our NFGD. Fig. \ref{['fig:location_distribution']} Glass surface location distribution. Fig. \ref{['fig:area_ratio']} Ratio of glass area against image area.
• Figure 5: The architecture of our proposed No-flash and Flash Glass Surface Detection Network (NFGlassNet).