Connie Ko is the GIS / Remote Sensing technician, adjunct faculty member and research associate at York University. She finished her doctorate from the Department of Earth and Space Science and Engineering at Lassonde School of Engineering, York University. Besides her work as a GIS / Remote Sensing technician for 18 years, which includes interaction with a wide range of students, faculty members, and staff, she is also involved in several research projects at York University. In particular, she is interested in the processing, analyzing and mapping of LiDAR (Light Detection and Ranging) data. Ko has 13 years of LiDAR data experience, and her current research interest involves the development and application of AI for 3D object detection using LiDAR data.

In the past few decades, the geomatics discipline has undergone revolutionary changes attributed to the advancement in many fields of science and engineering for sensing, positioning and navigation. Thanks to this technological advancement, a huge amount of data that we have never experienced before is available for the community to understand natural and built environments with much higher temporal, spectral and spatial dimensions. While the data acquisition systems are more accessible, the data quantity and complexity of the incoming data also increased. This challenge requires new theory, methods and system implements in remote sensing data analytics. Recently, deep learning (a sub-discipline in AI) has been demonstrating an enormous success in visual data analytics.

As part of the 3D Mobile Mapping Artificial Intelligence (3dmmai) project, she is interested in urban tree detection. Her study area is at the York University Keele Campus. Collaborating with Teledyne Optech, three sets of data is acquired for the project 1) Airborne LiDAR (Galaxy and Titan Sensor) 2) Mobile LiDAR (Lynx) and 3) field data. Figure 1 shows LiDAR point cloud acquired by Galaxy sensor on top of 3D tree inventory model which she generated for Intelligent Systems for Sustainable Urban Mobility (ISSUM) project at York Blvd at York University Keele Campus. Approximately 5700 individual trees is field validated on campus and use for training the AI network. Figure 2 shows one of the prediction results, green boxes are ground truth labeling and red boxes are predictions. This trained AI network is now able to predict tree locations in another geographical area where training has not been performed. In the summer of 2021, we tested the network on a power line right-of-way study site located in Steamboat Springs, Colorado, USA and obtained have obtained successful results.

Also under the 3dmmai project, she is now co-supervising Hyungju Lee for street furniture detection with mobile LiDAR (Lynx sensor). On top of trees on campus, they are now including classes such as bench, bus, bus shelter, car, garbage bin, light pole, pedestrian, road sign, traffic light, truck and utility pole. While these street furniture detections are traditionally not included in autonomous driving related research, they are essential items in many other applications such as creating a virtual reality twin.

Under ISSUM, Ko is collaborating with ESRI for creating a virtual York University Campus, including terrain height derived from Galaxy data, updated 3D tree inventory model (species specific), street furniture models, and detail textured 3D building exterior as well as some interior 3D floor plans (Figure 3). The campus model is currently supporting few other research such as psychological studies with navigation, and crowd simulation.