Xilei Zeng

About

Xilei Zeng is a pioneering researcher at the intersection of agricultural robotics and computer vision, whose work is revolutionizing autonomous fruit harvesting. His primary research areas include deep learning, 3D reconstruction, and real-time image segmentation for agricultural applications. Zeng’s major contributions lie in developing lightweight, high-accuracy neural networks that enable robots to perceive and navigate complex orchard environments. His most impactful work, the "lab-customized autonomous humanoid apple harvesting robot" (2021), has garnered 60 citations and demonstrates a complete system for fruit picking. He further advanced the field with "U2ESPNet" (2022, 22 citations), a model for real-time semantic segmentation of branches, and "MT-MVSNet" (2024, 10 citations), which uses a mobile transformer for efficient 3D reconstruction of fruit tree branches. Zeng’s innovative approach balances computational efficiency with accuracy, as seen in "ET-PatchNet" (2025, 6 citations) and "U-DPnet" (2023, 3 citations). His work on branch detection and reconstruction, including "Bunet" (2023, 2 citations), is critical for enabling robots to navigate without damaging trees. With over 140 total citations, Zeng is shaping the future of precision agriculture.

Research Focus

Key Achievements

Top Papers

1. 1
   A lab-customized autonomous humanoid apple harvesting robot

   60 citations · 2021
2. 2
   A real-time branch detection and reconstruction mechanism for harvesting robot via convolutional neural network and image segmentation

   43 citations · 2021
3. 3
   U2ESPNet—A lightweight and high-accuracy convolutional neural network for real-time semantic segmentation of visible branches

   22 citations · 2022
4. 4
   MT-MVSNet: A lightweight and highly accurate convolutional neural network based on mobile transformer for 3D reconstruction of orchard fruit tree branches

   10 citations · 2024
5. 5
   ET-PatchNet: A low-memory, efficient model for Multi-view Stereo with a case study on the 3D reconstruction of fruit tree branches

   6 citations · 2025
6. 6
   U-DPnet: an ultralight convolutional neural network for the detection of apples in orchards

   3 citations · 2023
7. 7
   Bunet: An effective and efficient segmentation method based on bilateral encoder-decoder structure for rapid detection of apple tree branches

   2 citations · 2023

Key Collaborators

Contact & Links