IP102: A Large-Scale Benchmark Dataset for Insect Pest Recognition

75,222 images of insect pests across 102 categories, preprocessed to 128x128 pixels. The dataset was created by Xiaoping Wu et al. for the IEEE CVPR conference in 2019 and later adapted for the Meta-Album benchmark in March 2022. It includes a hierarchical taxonomy grouping pests by the agricultural products they affect.

Use Cases

• Training image classifiers for insect pest recognition based on the 102 categories.
• Developing few-shot learning models using the preprocessed, standardized 128x128 images.
• Benchmarking model performance on a large-scale, domain-specific image dataset.
• Studying hierarchical classification based on the dataset's taxonomy grouping pests by crop.

Strengths

• Large scale with 75,222 images.
• High class diversity with 102 distinct insect pest categories.
• Images are preprocessed to a consistent 128x128 pixel size.
• Includes a hierarchical taxonomy for structured classification tasks.

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.

Provenance

Source

IP102: A Large-Scale Benchmark Dataset for Insect Pest Recognition (https://github.com/xpwu95/IP102)

Collection Method

Images were cropped into perfect squares and resized to 128x128 pixels for the Meta-Album release.

Time Range

Original dataset released in 2019.

Freshness

Created for Meta-Album on 01 March 2022; original data from 2019.

Geography

null