Abstract

Zoom link: https://cam-ac-uk.zoom.us/j/88158575361 Meeting ID: 881 5857 5361 Passcode: 277705

Abstract:

Trunk diameter is related to the overall health and level of carbon sequestration in a tree. Trunk diameter measurement, therefore, is a key task in both forest plot and urban settings. Unlike the traditional approach of manual measurement with a measuring tape or calipers, several recent approaches rely on sophisticated technologies such as LiDAR and time-of-flight cameras that provide fine-grain depth maps. These technologies are supported only on specialized devices or high-end smartphones. We present a mobile application called GreenLens that only uses coarse-grain depth maps derived from an optical sensor, and so can be run on most common Android devices. Moreover, we use a state-of-the-art deep neural network to estimate trunk diameter from an image and its corresponding coarse depth map (RGB-D). We tested our app under challenging conditions including occlusion, leaning trees, and irregular shapes and found that our algorithm is comparable to accuracy from fine-grain depth maps.

Currently, we are developing GreenLens2. Unlike our previous work, we are using a game engine (Unreal Engine) to create a highly photo-realistic virtual forest, making it easy to collect unlimited and diverse data for training neural networks. At the same time, we have proposed a multi-task neural network that performs trunk segmentation and end-to-end trunk diameter prediction simultaneously. We have also refined the app’s user journey to make it more interactive, straightforward, and user-friendly.

Bio:

Frank Feng is currently an undergraduate researcher and will join the Department of Computer Science and Technology at the University of Cambridge as a PhD student in October 2024.