Autonomous robots or unmanned aerial vehicles (UAV) can be used to acquire the necessary input (image/video) for yield estimation on centimetre scale. Difficulties come across when using those instruments in the field. One of the major challenges are the different light conditions, occlusion of the fruits, distance of camera to fruits, fruits clustering, etc.
The advantages of these platforms are that site-specific determination of yield can be done by combining real time imagery with deep learning to come to a near real time assessment of the yield. The real time component offers the farmers an immediate insight in the yield estimates of the orchard. Hence, it will accelerate the decision-making process.
This project aims to develop an approach for detecting and tracking apples in real-time using a UAV and deep learning. This will be achieved within the main steps:
1) Review of the state-of-the art
2) Design an approach
3) Experiments in available datasets
4) Writing report
This project will build on a previous thesis.
- Stepping towards real-time detection and tracking of apples using deep learning (MOTS). https://youtu.be/6pNhdK-T_e4
- Voigtlaender, P., Krause, M., Osep, A., Luiten, J., Sekar, B. B. G., Geiger, A., and Leibe, B. (2019). Mots: Multi-object tracking and segmentation. Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 2019-June:7934–7943.
- Häni, N., Roy, P., and Isler, V. (2020). A comparative study of fruit detection and counting methods for yield mapping in apple orchards. Journal of Field Robotics, 37(2):263–282.
- Enthusiast about drones/UAV
- Willing to learn more about real time image processing
- Excited to work in agriculture robotics
Theme(s): Sensing & measuring