New plant.health feature: follow-up questions

Our API now delivers a one of carefully formulated questions designed to differentiate between common plant health issues—ranging from water stress and nutrient deficiencies to pest infestations. For example, one of the questions in our assessment related to the photo of the poor peace lily below is:

Is the substrate moist?

For this question, the API returns a JSON response with specific diseases in the case of a positive or negative answer:

{
    "question": {
        "text": "Is the substrate moist?",
        "options": {
            "yes": {
                "suggestion_index": 0,
                "entity_id": "e5eed7f688efa59e",
                "name": "water excess or uneven watering"
            },
            "no": {
                "suggestion_index": 1,
                "entity_id": "c6fbc333b227bd54",
                "name": "water deficiency"
            }
        }
    }
}

In this example, if the user confirms that the substrate (soil) is moist, the API suggests water excess or uneven watering. Conversely, if the substrate is not moist, the answer is water deficiency. This simple, yes/no format helps to refine the model's response in cases where a multiple diseases show similar symptoms.

Here is an example of the implementation:

How is the question selected? First, the two most likely diseases based on the plant.health API are considered. Then a simple yet clever rule-based system then selects the best question that can distinguish between the two suggestions.

The question is now returned with every plant.health API call, making it easy to integrate this feature into your application.

You can try the functionality in the plant.id demo.