35th Annual Student Research Conference
Using Convolutional Neural Networks to Identify Snake Species
Samuel J. Myers
Dr. Ruthie Halma, Faculty Mentor
The goal of this study was to create an artificial intelligence (AI) capable of correctly identifying a snake species with accuracy greater than 90%. A secondary goal was analyzing the effect the number of epochs has on accuracy, where an epoch is the number of times the program cycles through the full training dataset. An AI was developed utilizing two convolutional neural networks with a variable number of epochs. This system was trained with 104 images per species, sized 384 x 384 pixels. Using this system, across 5 levels of epochs (50, 100, 150, 200, 250), three trials were run, with averaging accuracy rates of 69%, 88.33%, 89.66%, 94.33%, and 98.33%, respectively. Accuracy of 99% was achieved once during development, clearly showing that increasing the number of epochs also increases the accuracy of the system, albeit with diminishing returns.
Keywords: Artificial Intelligence, Machine Learning, Neural Networks, Computer Science, Animal Identification
Topic(s):Computer Science
Presentation Type: Oral Presentation
Session: 107-4
Location: MG 1098
Time: 9:15