IP Poster Session

[IPp-15] A Proposed Method for Estimating Pufferfish piece Counting and Pufferfish Size Using a Single Convolutional Neural Network

*Azu Murakami¹, Hiroki Takatsuka¹, Shiro Suyama¹, Masaki Yasugi^1,2, Hirotsugu Yamamoto¹ (1. Utsunomiya University, 2. Fukui Prefectural University)

We propose a method for estimating pufferfish piece counting and pufferfish size by using a single convolutional neural network. The piece counting has correct count in 56 images out of 100 images. Measurements were taken with respect to the size of the box in relation to the size of the fish. The most common bounding box size obtained from this measurement was between 0.05 and 0.1. This is 135 out of 344 total bounding boxes.