Automatic Flower Disease Identification Using Deep Convolutional Neural Network

Ashebr, Desalegn

dc.contributor.author	Ashebr, Desalegn
dc.date.accessioned	2020-06-04T07:00:05Z
dc.date.available	2020-06-04T07:00:05Z
dc.date.issued	2020-02
dc.identifier.uri	http://hdl.handle.net/123456789/10882
dc.description.abstract	Nowadays flower cultivation is the most predominant in Agricultural business. However, during the Cultivation of flowers, there are a number of risks that affect the flower, one of which is a disease. Fungus, bacteria, and viruses cause most flower diseases. Due to having large number of cultivated flowers, qualified agronomists and plant pathologists are strugglingto diagnose specific diseases correctly, therefore led to mistake conclusions and treatments. Identify the disease on the basis of manual feature extractions the result severally affected the overall outcomes. In image processing, feature extraction is timeconsuming and expensive which must be changed whenever the problem or the dataset changes, the recognition rate still fill to a local minimum and be contingent on the number of class. Primary analysis and exact credentials of flower diseases can rheostat the feast of contagion and ensure the well advance of the flower farming, Currently, various researcher uses multifaceted image processing and cannot assur high identification rates for flower diseases. This thesis suggests effective recognition approach for flower diseases using convolutional neural networks, It includes collecting adequatelye Damaged images of nine-flower disease and the health flower from those images, 70% of the Dataset is used for training and 15% used for validation and 15% used for testing. Next, build a convolutional neural network model used to identify and classify flower diseases. In this thesis convolutional neural network model is trained to identify the nine common flower diseases, with a test set. Image augmentation and Image Segmentation technique applied in this thesis in order to increase the performance of flower disease identification and computationally effective. Finally, after evaluated the model by using the test set we have achieved 92.41% test accuracy, 9.41% bettet than previous work.	en_US
dc.language.iso	en	en_US
dc.subject	Software Engineering	en_US
dc.title	Automatic Flower Disease Identification Using Deep Convolutional Neural Network	en_US
dc.type	Thesis	en_US