Title: View Planning for 3D Reconstruction of Plants
Speaker: Dr. Nikolaos Papanikolopoulos (University of Minnesota)
Time: Mar 19, 2026, 10:00 AM Eastern Time (9:00–10:00 AM Central Standard Time)
Zoom link:
https://binghamton.zoom.us/j/94071831448?pwd=bjAMGIdxfGy0X2Jf07Gc9syJreUCKd.1
Meeting ID: 940 7183 1448
Passcode: 598953
Abstract:
Active vision (AV) has been in the spotlight of robotics research due to its emergence in numerous applications, including agriculture and biomedicine. A major AV problem that has gained popularity is the 3D reconstruction of targeted environments from multiple 2D views. While collecting and processing a large number of arbitrarily taken 2D images may become an arduous process in several practical settings, an efficient solution is to seek the optimal placement of available cameras in the 3D space to obtain the necessary visual information from fewer yet more informative images to effectively reconstruct environments of interest. This process, termed as view planning (VP), can be markedly challenged in the presence of noise emerging in the environment, camera locations, and/or extracted images.
We present an efficient and realistic VP pipeline, which aims to optimize the viewpoints of cameras and hence the quality of the 3D reconstruction of a field of row crops without the need for a given mesh model. This is achieved within four steps: (i) an initial flight to obtain a sparse point cloud, (ii) the generation of an initial simple mesh model utilizing the sparse point cloud, (iii) the planning of images via a discrete optimization process, and (iv) a second flight to obtain the final reconstruction. We demonstrate the effectiveness of the proposed VP framework against commonly used baseline methods for agricultural data collection and processing.