Images from the COCO dataset filtered to include only the 'Person' class. The images have been cleaned and resized to facilitate faster model training. The specific number of images, rows, and columns is unknown.
Use Cases
- Train a single-class object detection model for the 'Person' class using the preprocessed images.
- Fine-tune a human detection model on cleaned and resized images for faster training cycles.
- Benchmark person detection algorithms on a curated subset of the COCO dataset.
- Use the 'Person' class images as a pretraining dataset for other computer vision tasks.
Strengths
- Derived from the widely used COCO dataset, a standard benchmark in computer vision.
- Images have been cleaned and resized, which reduces preprocessing overhead for users.
- Focus on a single 'Person' class simplifies tasks like human detection.
Limitations
- The dataset scope is limited to a single object class, which restricts multi-class applications.
- The specific number of images and details of the cleaning process are unknown.
- Without the original COCO annotations, certain types of analysis may be impossible.
Provenance
- Source
- COCO (Common Objects in Context) dataset.
- Collection Method
- Filtered, cleaned, and resized from the original COCO dataset.
- Time Range
- null
- Freshness
- null
- Geography
- null