Farmers Can Check Crop Problems With a Phone. How Reliable Are AI Disease Diagnosis Tools?

A farmer noticing spots, yellow leaves or damaged plants often has to identify the problem quickly. Waiting for an expert visit can delay treatment. Smartphone-based artificial intelligence tools are now being developed to help farmers identify crop diseases, pests and other stresses by simply taking a photograph.

India has several such systems. ICAR-IASRI’s AI-DISC can identify diseases and insect pests using uploaded crop images and also connects users with experts. The RAISE app developed around ICAR-Indian Institute of Rice Research can identify several rice stresses, including pests, diseases, weeds, drought, salinity and nutrient deficiencies.

The technology is also moving beyond research projects. FasalNetra, developed by the Annam.AI Centre of Excellence at IIT Ropar, uses computer vision to identify crop pests and diseases and provides confidence scores, severity assessment and management recommendations.

These tools do not replace agricultural scientists. Their main value is helping farmers recognise a possible problem quickly and decide whether they need further expert support.

How does image-based diagnosis work?

AI disease-detection tools are trained using large collections of crop images. The system learns visual patterns associated with particular diseases, pests or stresses. When a farmer uploads a new photograph, the model compares its features with patterns learned during training and produces a possible diagnosis.

RAISE was developed specifically for rice. Its current application reports approximately 99% accuracy on its training dataset and 85–90% when validated using field-acquired images. The difference is important because real farms contain changing light, backgrounds, varieties and symptoms that can be harder to classify.

The app can examine photographs and provide an immediate result. ICAR-IIRR says the system is intended to make it easier for farmers to identify problems and receive suitable management recommendations.

AI-DISC takes another approach by combining automated identification with an expert forum. Farmers can upload photographs and metadata, while submitted images can be validated by domain experts. This creates a feedback system in which field observations can also contribute to a wider repository of crop disease and pest images.

This combination of AI and human review is important because similar symptoms can have different causes. Yellowing, leaf spots or poor growth may result from disease, insects, nutrient deficiency, water stress or several problems occurring together.

Can AI tell farmers which pesticide to use?

Some applications provide management recommendations after identifying a likely problem. But farmers should not treat an AI result as an automatic pesticide prescription.

The National Pest Surveillance System, developed under ICAR, combines AI-based pest identification with pest reporting, surveillance and location-specific advisories. ICAR says the system supports real-time pest detection and can help deliver timely crop protection advice through the KVK network.

The system covers 66 crops and more than 432 pest types, according to the Agriculture Ministry. It provides information to extension workers who can use the data to support farmers. The official NPSS mobile application was updated in September 2026.

This approach can be safer than relying only on a photograph. An AI system can suggest what a symptom resembles, while agricultural experts can consider crop stage, pest population, weather, local outbreaks and economic thresholds before recommending control measures.

Farmers should therefore avoid spraying a pesticide simply because an application identifies a possible pest or disease. The diagnosis should be checked, especially when the result is uncertain or the recommended treatment involves a chemical pesticide.

What should farmers do when using these apps?

The quality of the photograph matters. Farmers should take clear pictures in good light and capture the affected part of the plant as well as the whole plant when possible. Blurry photographs, poor lighting or a very small affected area can make identification more difficult.

Farmers should also provide useful information when an application asks for it. Crop name, variety, location, crop stage and recent weather or management details can help an advisory system produce a more useful response.

RAISE has also been designed with future geo-tagging in mind. ICAR-IIRR said location information from uploaded images could help identify areas where particular pests or diseases are appearing more frequently and support early warning alerts.

Another ICAR tool, riceXpert, developed by ICAR-NRRI, allows farmers to send text, photographs and recorded voice messages about rice problems. Farm scientists can respond with solutions through the system. ICAR-NRRI reported around 5,000 users and about 145 farmer queries addressed through the platform as of August 2026.

The growing number of tools shows that AI-based crop diagnosis is becoming more accessible. But farmers should use it as an early-warning and decision-support tool, not as a replacement for field observation or expert advice.

The biggest benefit may come when the two work together. A farmer can photograph a suspected problem, receive a preliminary diagnosis and then share the result with a KVK or agricultural expert. This can save time while reducing the risk of acting on an incorrect diagnosis.