Plastic Debris Images from Controlled Marine Tank Experiments

4,511 images of plastic debris were captured in a tank at the University of Washington's Ocean Technology Center using a 1080p webcam. The dataset was created by OceanCV and uploaded to Hugging Face on November 4, 2024. Its goal is to enhance the detection and classification of plastic types commonly found in marine environments.

Use Cases

• Training object detection models based on images of plastic debris in water.
• Developing classification models for different types of plastic based on visual features.
• Benchmarking model performance for marine debris detection in controlled environments.
• Creating educational or demonstration tools for environmental research based on the described imagery.

Strengths

• Contains 4,511 images, providing a substantial volume for model training.
• Images were captured in a controlled tank environment, likely ensuring consistent lighting and background conditions.
• Data originates from a university research center, suggesting a scientific collection method.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Data may reflect bias inherent to the controlled tank setup, not representing real-world marine conditions.

Provenance

Source

OceanCV, likely affiliated with the University of Washington's Ocean Technology Center.

Collection Method

Images were captured using a 1080p webcam in a tank containing different types of plastic.

Freshness

Last updated 2024-11-04 21:55:52; freshness should be verified.

Geography

Data was collected at the University of Washington's Ocean Technology Center; specific geographic coverage is unknown.