ML Based Smart Silkworm Egg Count System For Sericulture

Sericulture is an important agricultural activity in India and is mainly carried out in rural areas. In this field, silkworm eggs are supplied to farmers in the form of Disease- Free Laying (DFL) sheets. Normally, the counting of eggs and checking whether the eggs are hatched or not is done manually in grainage centers. This process requires a lot of time and depends on human, which can always cause the mistakes, especially when the number of eggs is large.In this project, a Machine Learning based system was developed to count silkworm eggs and also to check whether the eggs are hatched or not. Images of DFL sheets were collected during the project work and used as input to the system then the Basic image processing steps were an applied mainly to improve the visibility, although in some images the improvement was not very clear. A Convolutional Neural Network (CNN) model was then used for egg classification, using the image data that was collected during testing. During testing, it was noticed that the system worked better when the images were the clear and a properly captured. For low-light images, the output was not always the same and small differences were noticed during testing. In most tests, the accuracy a stayed around 90% to 93%. This was taken as acceptable the for this project work, since the main aim was to the reduce a manual effort rather than achieve perfect results.