Research Projects

The project aims to develop a non-invasive ultrasound technique for detecting the sex, assessing the gonadal maturity stages of both male and female fishes of brackishwater aquaculture importance. The findings will help in obtaining stress free spawning without any obstruction in the reproductive success of captive brooders. The technique to be developed will be of great help for private entrepreneurs to make use of user-friendly technique which will unblock the hurdles to have a strengthened scientific knowledge and facilities to start up the fin fish hatchery especially the foremost frontier of brooders development and breeding. The stress- free spawning is achieved through deep learning techniques. The ultrasound images are acquired and preprocessed using windowing, segmentation and feature extraction to remove the speckle noise which extracts texture features, shape features, histogram, correlogram features and morphology features. The processed image is fed to the deep learning technique to identify the gonads stages which are classified into immature, mature, ripe and running. Transfer learning is used to populate more images for training the system to provide better accuracy. The Generative Adversarial Network (GAN) is used to classify the gonad stages. Finally, the stress-free deep learning results are compared with the histology data for validating the test results