<![CDATA[There was a spike in the ornamental plants as a hobby while spending time at home during the COVID pandemic when people were restricted to activities outside the house. Unfortunately, along with this trend also came the serious issues associated with fake reports claiming that some ornamental plants were harmful to people's health. The public is more worried and perplexed by this situation, which also erodes their confidence in ornamental plants. This research aims to develop a real-time ornamental plant identification system as an educational medium for the public. To increase the system's accuracy, the transfer learning method is applied to the modified MobileNet CNN model. There are 9 species of popular ornamental plants in this identification system. From the experiments, it is known that the best accuracy has been achieved using the Adagrad optimization method (96% for training and 88% for testing data). The CNN model is then embedded in PLANTIS, an Android-based application prototype for ease of use purpose.]]>
Copyrights © 2024
