Francesco Caccivio

BEng Industrial Engineering
BEng Peirianneg Ddiwydiannol

Deep learning analysis with convolutional neuronal networks.

Dadansoddiad dysgu dwfn gyda rhwydweithiau niwronau troellol.

Deep Learning is ubiquitous and opens doors to interesting projects. It should be possible to realise such a project with minimal resources. The project should be carried out from the creation of a data set to the modelling of a neural network and the testing of the model. The data set was created using a Raspberry Pi Camera. It contains images of straight roads, left-hand curves, and right-hand curves. The model was developed based on a convolutional neural network. It has an accuracy of 97% on new data.

Mae Dysgu Dwfn yn hollbresennol ac yn agor drysau i brosiectau diddorol. Dylai fod yn bosibl gwireddu prosiect o'r fath heb fawr o adnoddau. Dylid cynnal y prosiect o greu set ddata i fodelu rhwydwaith niwrol a phrofi'r model. Crëwyd y set ddata gan ddefnyddio Camera Raspberry Pi. Mae'n cynnwys delweddau o ffyrdd syth, troadau i’r chwith, a throadau i’r dde. Datblygwyd y model yn seiliedig ar rwydwaith niwral troellol. Mae ganddo gywirdeb o 97% ar ddata newydd.