IndoorOutdoorNet-20K: 19,998 Labeled Images for Indoor-Outdoor Classification

IndoorOutdoorNet-20K is a labeled image dataset containing 19,998 images for binary classification between indoor and outdoor scenes. It was created by prithivMLmods and uploaded to Hugging Face Datasets on 2025-04-22. The dataset is intended for tasks like scene understanding, transfer learning, and model benchmarking.

Use Cases

• Train binary image classifiers based on the indoor/outdoor scene labels.
• Benchmark model performance on scene classification tasks based on the dataset's defined task.
• Apply transfer learning from models trained on this dataset to other visual recognition tasks based on its image modality.

Strengths

• Contains 19,998 labeled images, providing a substantial volume for training.
• Focuses on a clear binary classification task (Indoor vs. Outdoor) for scene understanding.

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 dataset split is reported as 100% train, which may limit its use for validation or testing without further processing.

Provenance

Source

huggingface

Freshness

Last updated 2025-04-22 16:49:39; freshness should be verified.