Sign in to view source links and access this dataset
Description
2,710 training images with full annotations for glass surface detection in the wild, as described in the CVPR 2021 paper 'Rich Context Aggregation with Reflection Prior for Glass Surface Detection'. The dataset was created by Jiaying Lin, Zebang He, and Rynson W.H. Lau and is hosted by the author 'garrying' on Hugging Face.
Use Cases
Training glass surface segmentation models based on the provided mask annotations.
Analyzing reflection properties in real-world scenes based on the reflection annotations.
Developing edge detection algorithms for transparent objects based on the edge annotations.
Benchmarking model performance for the specific task of glass surface detection.
Strengths
2,710 annotated images in the training split provide a substantial base for model development.
Annotations include masks, reflection data, and edges, offering multi-faceted labels for a single task.
Limitations
Description metadata is limited; actual data quality requires manual inspection after download.
Column-level documentation is absent; field semantics must be inferred after download.
The total dataset size, including validation and test splits, is unknown.
Provenance
Source
CVPR 2021 paper 'Rich Context Aggregation with Reflection Prior for Glass Surface Detection'.
Collection Method
Likely collected and annotated for the specific research project.
Time Range
null
Freshness
Last updated 2026-04-30 17:41:51; freshness should be verified.
Geography
null
License is unknown; terms of use must be verified before application.