Edge AI-Driven System for Silkworm Pupae Gender Classification and Process Control

In India, silkworms are essential to the production of silk. Silkworms have a 6–8 week life cycles. The process of turning a silkworm into a mature Bombyx mori moth yields peace of silk. Silkworms are more valuable because of their nutritional makeup, which includes vitamins, oils, proteins, minerals, and other beneficial elements. Silkworm diseases constitute a significant threat to the sericulture industry, causing substantial economic losses. However, there were research gaps in the areas of environmental factors and disease outbreaks. Existing diagnostic tools are limited in understanding complex disease interactions related to silkworm rearing. Addressing these research gaps is essential for improving silkworm disease prevention, control, and overall sericulture sustainability. Visual and time-consuming classification methods lead to errors, and it reduces productivity. Transfer learning was implemented for image classification. Gender classification of silkworm pupae using ResNetV2 architecture achieved 97% accuracy. The development of an automated system is done by employing a robotic arm integrated with a camera, Raspberry Pi, and Arduino Uno. A camera coupled with a Raspberry Pi captures high-resolution images of the pupae, subsequently processed for classification. Arduino Uno controls the robotic arm’s movements to classify each silkworm pupa by the category, like male, female or diseased. This automated AI based embedded hardware efficiently classifies the gender of silkworm pupa and hence reduces human intervention. The quality of silk production will be improved.