Search

Abstract Background The study of plant photosynthesis is essential for productivity and yield. Thanks to the development of high-throughput phenotyping (HTP) facilities, based on chlorophyll fluorescence imaging, photosynthetic traits can be measured in a reliable, reproducible and efficient manner. In most state-of-the-art HTP platforms, these traits are automatedly analyzed at individual plant level, but information at leaf level is often restricted by the use of manual annotation. Automated leaf tracking over time is therefore highly desired. Methods for tracking individual leaves are still uncommon, convoluted, or require large datasets. Hence, applications and libraries with different techniques are required. New phenotyping platforms are initiated now more frequently than ever; however, the application of advanced computer vision techniques, such as convolutional neural networks, is still growing at a slow pace. Here, we provide a method for leaf segmentation and tracking through the fine-tuning of Mask R-CNN and intersection over union as a solution for leaf tracking on top-down images of plants. We also provide datasets and code for training and testing on both detection and tracking of individual leaves, aiming to stimulate the community to expand the current methodologies on this topic. Results We tested the results for detection and segmentation on 523 Arabidopsis thaliana leaves at three different stages of development from which we obtained a mean F-score of 0.956 on detection and 0.844 on segmentation overlap through the intersection over union (IoU). On the tracking side, we tested nine different plants with 191 leaves. A total of 161 leaves were tracked without issues, accounting to a total of 84.29% correct tracking, and a Higher Order Tracking Accuracy (HOTA) of 0.846. In our case study, leaf age and leaf order influenced photosynthetic capacity and photosynthetic response to light treatments. Leaf-dependent photosynthesis varies according to the genetic background. Conclusion The method provided is robust for leaf tracking on top-down images. Although one of the strong components of the method is the low requirement in training data to achieve a good base result (based on fine-tuning), most of the tracking issues found could be solved by expanding the training dataset for the Mask R-CNN model.

Fusarium head blight (FHB) is a plant disease caused by various species of the Fusarium fungus. One of the major concerns associated with Fusarium spp. is their ability to produce mycotoxins. Mycotoxin contamination in small grain cereals is a risk to human and animal health and leads to major economic losses. A reliable site-specific precise Fusarium spp. infection early warning model is, therefore, needed to ensure food and feed safety by the early detection of contamination hotspots, enabling effective and efficient fungicide applications, and providing FHB prevention management advice. Such precision farming techniques contribute to environmentally friendly production and sustainable agriculture. This study developed a predictive model, Sága, for on-site FHB detection in wheat using imaging spectroscopy and deep learning. Data were collected from an experimental field in 2021 including (1) an experimental field inoculated with Fusarium spp. (52.5 m × 3 m) and (2) a control field (52.5 m × 3 m) not inoculated with Fusarium spp. and sprayed with fungicides. Imaging spectroscopy data (hyperspectral images) were collected from both the experimental and control fields with the ground truth of Fusarium-infected ear and healthy ear, respectively. Deep learning approaches (pretrained YOLOv5 and DeepMAC on Global Wheat Head Detection (GWHD) dataset) were used to segment wheat ears and XGBoost was used to analyze the hyperspectral information related to the wheat ears and make predictions of Fusarium-infected wheat ear and healthy wheat ear. The results showed that deep learning methods can automatically detect and segment the ears of wheat by applying pretrained models. The predictive model can accurately detect infected areas in a wheat field, achieving mean accuracy and F1 scores exceeding 89%. The proposed model, Sága, could facilitate the early detection of Fusarium spp. to increase the fungicide use efficiency and limit mycotoxin contamination.

A phenotyping pipeline utilising DeepLab was developed for precisely estimating the height, volume, coverage and vegetation indices of European and Japanese varieties. Using this pipeline, the effect of varying UAV height on the precise estimation of potato crop growth properties was evaluated. A UAV fitted with a multispectral camera was flown at a height of 15 m and 30 m in an experimental field where various varieties of potatoes were grown. The properties of plant height, volume and NDVI were evaluated and compared with the manually obtained parameters. Strong linear correlations with R2 of 0.803 and 0.745 were obtained between the UAV obtained plant heights and manually estimated plant height when the UAV was flown at 15 m and 30 m respectively. Furthermore, high linear correlations with an R2 of 0.839 and 0.754 were obtained between the UAV-estimated volume and manually estimated volume when the UAV was flown at 15 m and 30 m respectively. For the vegetation indices, there were no observable differences in the NDVI values obtained from the UAV flown at the two heights. Furthermore, high linear correlations with R2 of 0.930 and 0.931 were obtained between UAV-estimated and manually measured NDVI at 15 m and 30 m respectively. It was found that UAV flown at the lower height had a higher ground sampling distance thus increased resolution leading to more precise estimation of both the height and volume of crops. For vegetation indices, flying the UAV at a higher height had no effect on the precision of NDVI estimates.

Accurately detecting and counting fruits during plant growth using imaging and computer vision is of importance not only from the point of view of reducing labor intensive manual measurements of phenotypic information, but also because it is a critical step toward automating processes such as harvesting. Deep learning based methods have emerged as the state-of-the-art techniques in many problems in image segmentation and classification, and have a lot of promise in challenging domains such as agriculture, where they can deal with the large variability in data better than classical computer vision methods. This paper reports results on the detection of tomatoes in images taken in a greenhouse, using the MaskRCNN algorithm, which detects objects and also the pixels corresponding to each object. Our experimental results on the detection of tomatoes from images taken in greenhouses using a RealSense camera are comparable to or better than the metrics reported by earlier work, even though those were obtained in laboratory conditions or using higher resolution images. Our results also show that MaskRCNN can implicitly learn object depth, which is necessary for background elimination.

Hyperspectral imaging is currently a very well-known and much used technology for measuring features in different fields, such as chemistry, geology, medicine, food and agriculture, either spaceborne (satellites), airborne (drones) or at close proximity (e.g. field scanning, industrial sorting lines or microscopy). Its background is two-fold, and it can be considered as a special case of spectroscopy (“imaging spectroscopy”) or a special case of imaging (“spectral imaging”). Current practice is to use adjectives such as multi and hyper added to “spectral imaging” in order to characterise the number of wavelength bands. In this paper we propose the community to use scientifically sound terminology, like “imaging spectroscopy” or “spectral imaging”, without using ambiguous adjectives. Further, we encourage the community to define and agree upon clear adjectives to describe the number of variables in the naming of our imaging technique.

High-throughput phenotyping is playing an increasingly important role in many areas of agriculture. Breeders will use it to obtain values for the traits of interest so that they can estimate genetic value and select promising varieties; growers may be interested in having predictions of yield well in advance of the actual harvest. In most phenotyping applications, image analysis plays an important role, drastically reducing the dependence on manual labor while being non-destructive. An automatic phenotyping system combines a reliable acquisition system, a high-performance segmentation algorithm for detecting fruits in individual images, and a registration algorithm that brings the images (and the corresponding detected plants or plant components) into a coherent spatial reference frame. Recently, significant advances have been made in the fields of robotics, image registration, and especially image segmentation, which each individually have improved the prospect of developing a fully integrated automatic phenotyping system. However, so far no complete phenotyping systems have been reported for routine use in a production environment. This work catalogs the outstanding issues that remain to be resolved by describing a prototype phenotyping system for a production tomato greenhouse, for many reasons a challenging environment.

Virus diseases are of high concern in the cultivation of seed potatoes. Once found in the field, virus diseased plants lead to declassification or even rejection of the seed lots resulting in a financial loss. Farmers put in a lot of effort to detect diseased plants and remove virus-diseased plants from the field. Nevertheless, dependent on the cultivar, virus diseased plants can be missed during visual observations in particular in an early stage of cultivation. Therefore, there is a need for fast and objective disease detection. Early detection of diseased plants with modern vision techniques can significantly reduce costs. Laboratory experiments in previous years showed that hyperspectral imaging clearly could distinguish healthy from virus infected potato plants. This paper reports on our first real field experiment. A new imaging setup was designed, consisting of a hyperspectral line-scan camera. Hyperspectral images were taken in the field with a line interval of 5 mm. A fully convolutional neural network was adapted for hyperspectral images and trained on two experimental rows in the field. The trained network was validated on two other rows, with different potato cultivars. For three of the four row/date combinations the precision and recall compared to conventional disease assessment exceeded 0.78 and 0.88, respectively. This proves the suitability of this method for real world disease detection.

Leaf area measurements are commonly obtained by destructive and laborious practice. This study shows how stereo and time‐of‐flight (ToF) images can be combined for non‐destructive automatic leaf area measurements. The authors focus on some challenging plant images captured in a greenhouse environment, and show that even the state‐of‐the‐art stereo methods produce unsatisfactory results. By transforming depth information in a ToF image to a localised search range for dense stereo, a global optimisation strategy is adopted for producing smooth results that preserve discontinuity. They also use edges of colour and disparity images for automatic leaf detection and develop a smoothing method necessary for accurately estimating surface area. In addition to show that combining stereo and ToF images gives superior qualitative and quantitative results, 149 automatic measurements on leaf area using the authors system in a validation trial have a correlation of 0.97 with true values and the root‐mean‐square error is 10.97 cm2, which is 9.3% of the average leaf area. Their approach could potentially be applied for combining other modalities of images with large difference in image resolutions and camera positions.