AMULET can use artificial intelligence to predict plant development. It can be used by farmers and scientists

Phenotyping line for non-invasive monitoring of plant traits in relation to the environment.
Source: ICOK
Wednesday 4 June 2025, 9:00 – Text: Martina Šaradínová

AMULET enables efficient and accurate measurement of morphological and physiological characteristics of plants. Plant research experts and computer scientists from CATRIN at Palacký University, VŠB-Technical University Ostrava and Imperial College London collaborated in its development and presented the results in the journal Computers and Electronics in Agriculture. By combining a range of imaging methods with advanced machine learning algorithms, the system can predict the future development and condition of plants based on the data obtained, which is crucial information for farmers and crop breeders.

“In this study, we have shown that AMULET can significantly improve the process of phenotyping data, i.e. monitoring and evaluating plant traits in relation to the environment. This is crucial for plant breeding and agricultural research. This approach allows faster and more accurate analysis of plant traits, which can contribute to the development of more resistant and profitable crops,” said Nuria De Diego, one of the authors.

The AMULET model processes images acquired using an affordable RGB camera. The researchers have trained it on more than 30,000 images of a model plant, the Arabidopsis thaliana, but have shown that it can also be used on agricultural crops such as potatoes.

 “The system includes plant detection, estimation of future development, sorting and data analysis. It improves phenotyping by using an advanced artificial intelligence model that can predict the evolution of image data with high accuracy. This capability benefits a wide range of users – from scientists to farmers – for example, by shortening the duration of experiments, enabling early detection of plant stress or faster identification of unhealthy individuals,” explained the paper’s first author Jan Zdražil from CATRIN and a PhD student at the Faculty of Electrical Engineering and Power Engineering at VŠB-TUO.

The researchers have also tested AMULET in plants that had “encountered” the pathogen Pseudomonas syringae. AMULET was also able to detect plant traits that are difficult for the human eye to perceive, but contribute significantly to understanding the response of plants to specific growth conditions.

“AMULET has also proven itself in predicting disease onset in plants before the first visible symptoms appear, allowing early intervention and minimising yield loss. This will allow for faster and more responsive intervention to protect crop health more effectively,” added De Diego.

AMULET was developed with the support of the European project PATAFEST from the Horizon Europe RIA programme.  Although experts say the system’s functionality still needs to be further tested in a wide range of conditions and plant species, it is already a breakthrough tool that can fundamentally improve phenotyping – from detection to data analysis. If data can be obtained from the field to validate the model, its use in breeding programmes and agriculture may in the foreseeable future contribute to higher crop vigour and yield and enable proactive plant care with less time and labour.

Back

Privacy settings

We use cookies and any other network identifiers on our website that may contain personal data (e.g. about how you browse our website). We and some of the service providers we use have access to or store this data on your device. This data helps us to operate and improve our services. For some purposes, your consent is required to process data collected in this way. You can change or revoke your consent at any time (see the link at the bottom the page).

(Essential cookies enable basic functions and are necessary for the website to function properly.)
(Statistics cookies collect information anonymously. This information helps us to understand how our visitors use our website.)
(They are designed for promotional purposes, measuring the success of promotional campaigns, etc.)