IkanMu : Fish Species and Freshness Detection Application Using SSD MobileNet V3
Final Project at Orbit Future Academy , : Ikanmu mobile applicatio
The datasets used in this final project are images of fish obtained by taking approximately 300 images of each type of fish taken at TPI (Fish Auction Place) Cilacap and datasets from websites that provide fish datasets as additional datasets. Photographs were taken of skipjack and baby tuna species at TPI Cilacap every morning for two weeks.
Data from data retrieval goes through the data preprocessing stage which has been cut, cleaned, and multiplied data to get the best quality image. The cutting process is carried out with a size of 480x480 pixels to take pictures of the fish as a whole. Image selection is done on images of fish that are not intact. Multiply the image of the cut fish to get fish from various angles. The data preprocessing process produces 1,900 images that will be used for learning the SSD Mobilenet v3 algorithm. The images are divided by a ratio of 80:20 with 1,520 images for training and another 380 images used for model evaluation.