🌽 Maize
Initial targets can include fall armyworm damage, leaf blight, rust, maize streak symptoms and nutrient deficiencies.
Crop coverage
Ten crops define the first diagnostic library.
The initial catalogue focuses training, validation and field testing on crops widely grown in Nigeria. Each crop can contain disease, pest, nutrient-deficiency and visible-stress classes.
🌿 Cassava
Initial targets can include cassava mosaic symptoms, bacterial disease, mite damage and visible nutrient or water stress.
🌾 Sorghum
Leaf diseases, anthracnose-like symptoms, smuts, pests and deficiency patterns.
🫘 Cowpea
Leaf spots, mosaic-like symptoms, aphid damage, pod damage and visible nutrient stress.
🌾 Rice
Blast, bacterial leaf blight, brown spot and nutrient-deficiency patterns.
🌾 Millet
Downy mildew, smut, ergot-like symptoms and pest damage.
🍠 Yam
Anthracnose, viral symptoms, leaf spots and insect damage.
🥜 Groundnut
Early and late leaf spots, rust, rosette-like symptoms and nutrient stress.
🌱 Cocoyam
Leaf blight, leaf spots and visible stress or deficiency patterns.
🌱 Sesame
Leaf spot, phyllody-like symptoms, wilt and visible pest damage.
Field data
Real farm conditions are part of the training problem.
Useful vision models need images with sunlight, shadows, overlapping leaves, different crop stages, low-cost cameras and mixed damage patterns. CropVox is structured around field validation rather than clean leaf images alone.
Maize
Outdoor images help expose lighting and background variation.
Cassava
Whole-plant and leaf-level views can both matter.
Variable environments
Natural backgrounds make validation more realistic.