In this project, an investigation of a neural network based system is used to examine
the following: a) the possibility and practicability of analysing and recognising parasites/sealworms on a parasite/sealworm infested cod fish images, b) the most efficient but robust way of presenting data to the neural network for efficient training and generalisation. The basic problem is to automate the sorting of sealworm infested cod fish from good normal cod fish using a neural network based system. The generalised back propagation supervised learning algorithm is used and both steepest descent and conjugate gradient methods are investigated. Various data representation schemes in unprocessed and processed formats before presentation for training of the neural network, are also examined. Finally the level of recognition achieved by the neural network when presented with the cod fish images is computed. Thus in this project an attempt is made to analyse and find the best components for solving the basic problem and then use this information to develop a neural network based system to recognise, detect and locate parasite/sealworms on cod fish images. / Thesis / Master of Science (MSc)
| Identifer | oai:union.ndltd.org:mcmaster.ca/oai:macsphere.mcmaster.ca:11375/24508 |
| Date | 04 1900 |
| Creators | Aryee, Emmanuel |
| Contributors | Poehlman, S., Computation |
| Source Sets | McMaster University |
| Language | English |
| Detected Language | English |
| Type | Thesis |
Page generated in 0.0027 seconds