This dataset presents curated and annotated 256\u00d7256 pixel image fragments of thoracic and elytral regions from six economically significant species within the beetle families Curculionidae and Tenebrionidae. These anatomical fragments were extracted from high-resolution trap images to support numerical signature generation for species-level classification. The dataset enables researchers to develop and validate image-based machine learning pipelines for pest identification in stored product environments. All samples are standardized and include metadata indicating species, anatomical region, and imaging parameters. This work contributes to advancing computational entomology, post-harvest pest management, and real-time automated monitoring systems.
Insect pest detection in post-harvest storage environments is crucial for mitigating economic losses and ensuring food security. Traditional identification approaches are labor-intensive and rely on full-body morphological features. This dataset introduces a fragment-based classification framework using annotated thoracic and elytral sections of stored-product beetles. Species included are:
Folder abbreviations for the insects are Sitophilus zeamais (mw); Sitophilus oryzae (rw); Sitophilus granarius (ww); Tribolium castaneum (tc); Latheticus oryzae (lo); and Tribolium confusum (tco).
Key numerical signature descriptors were computed for each fragment: