Open World Dense Object Detection Dataset with 8,001 Images

A large-scale dataset for open-world object detection contains 8,001 images annotated with 344,079 bounding boxes across 1,106 categories. It was created by author shubh303 and last updated on January 5, 2026. The dataset is designed specifically for training and evaluating text-conditioned, open-vocabulary detection models.

Use Cases

• Training open-world object detection models based on the 1,106 diverse categories.
• Benchmarking text-conditioned detectors based on bounding box annotations.
• Evaluating model performance on dense, open-vocabulary scenarios based on the dataset's stated purpose.
• Researching the generalization capabilities of detection systems beyond closed-set training.

Strengths

• Dataset contains 8,001 images, providing a substantial collection for training.
• Annotations include 344,079 bounding boxes, indicating dense object labeling.
• Categories span 1,106 labels, supporting open-world and open-vocabulary research.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Freshness should be verified as the last update date is 2026-01-05.

Provenance

Source

huggingface

Freshness

Last updated 2026-01-05 07:00:48.