Real-Time Plant Leaf Counting Using Deep Object Detection Networks

Abstract

The use of deep neural networks (DNNs) in plant phenotyping has recently received considerable attention. By using DNNs, valuable insights into plant traits can be readily achieved. While these networks have made considerable advances in plant phenotyping, the results are processed too slowly to allow for real-time decision-making. Therefore, being able to perform plant phenotyping computations in real-time has become a critical part of precision agriculture and agricultural informatics. In this work, we utilize state-of-the-art object detection networks to accurately detect, count, and localize plant leaves in real-time. Our work includes the creation of an annotated dataset of Arabidopsis plants captured using Cannon Rebel XS camera. These images and annotations have been complied and made publicly available. This dataset is then fed into a Tiny-YOLOv3 network for training. The Tiny-YOLOv3 network is then able to converge and accurately perform real-time localization and counting of the leaves. We also create a simple robotics platform based on an Android phone and iRobot create2 to demonstrate the real-time capabilities of the network in the greenhouse. Additionally, a performance comparison is conducted between Tiny-YOLOv3 and Faster R-CNN. Unlike Tiny-YOLOv3, which is a single network that does localization and identification in a single pass, the Faster R-CNN network requires two steps to do localization and identification. While with Tiny-YOLOv3, inference time, F1 Score, and false positive rate (FPR) are improved compared to Faster R-CNN, other measures such as difference in count (DiC) and AP are worsened. Specifically, for our implementation of Tiny-YOLOv3, the inference time is under 0.01 s, the F1 Score is over 0.94, and the FPR is around 24%. Last, transfer learning using Tiny-YOLOv3 to detect larger leaves on a model trained only on smaller leaves is implemented. The main contributions of the paper are in creating dataset (shared with the research community), as well as the trained Tiny-YOLOv3 network for leaf localization and counting.

Automation of leaf counting in maize and sorghum using deep learning

“…The counting-by-regression approach is sensitive to the number of images representing each potential number of leaves a plant can have, with biased performance for rarer and more extreme leaf numbers (Figure 3a, Supplemental Figure S4). An alternative approach to leaf counting is to implement counting-by-detection models (Buzzy et al, 2020;Xu et al, 2018) built on top of object detection frameworks (Redmon et al, 2016;Ren et al, 2015). Compared to a single number as the label for each training image in building CNNs, Object detection models are trained using annotations of bounding boxes around objects of interest.…”

Robot navigation in vineyards based on the visual vanish point concept

“…With the goal of counting objects, Vasconez et al (2020) and Buzzy et al (2020) use object detection to identify and count. The first, detects and counts a variety of fruits (avocado, apple and lemons), in Fig.…”

Recent developments and potential of robotics in plant eco-phenotyping

“…has further advanced the ability to enhance the complete study of plant for ecophenotyping. In addition, Artificial Intelligence (AI) technologies, such as deep learning, big data mining, and machine learning, provide the means to process and interpret plant data which was collected via high-throughput devices [13][14][15][16].…”