Fish farmers relied on morning visual checks and necropsy after death. By the time symptoms were obvious, the disease had often spread to 30–50% of the tank. Treatment was reactive, not preventive.
How Fish Disease Was Diagnosed Traditionally
Visual inspection only catches late‑stage disease. Microscopy requires expertise and time. By then, Aeromonas ulcers and columnaris have already spread.
How AI Detects Fish Diseases
AI models are trained on thousands of labeled fish images — healthy, flashing, clamped fins, white spots, ulcers. Cameras capture real‑time video, and the model flags anomalies. The system alerts you to the first flash or fin clamp, often 48 hours before visible lesions appear.
Types of Fish Diseases AI Can Identify
Ectoparasites: Ich, velvet, trichodina (by flashing behavior and spot patterns)
Bacterial: Columnaris saddle lesions, Aeromonas red ulcers, fin rot
Viral: KHV gill necrosis, lymphocystis cauliflower growths
Behavioral: Dropsy pineconing, whirling disease spiraling, costia mucus film
Cameras and Imaging Equipment Used
Underwater 4K IP camera (e.g., Axis, Hikvision) — Mounted at 45° for side profile.
Raspberry Pi HQ camera + housing — DIY open‑source detection system.
Thermal camera — Detects gill inflammation before visual signs.
Hyperspectral imaging — Lab‑grade, detects pathogen metabolites in water.
Future of Smart Aquaculture
AI + water sensors combined — Model correlates low DO spike with subsequent ich outbreak.
Automated treatment response — System detects flashing → triggers formalin dip → logs event.
Mobile app diagnosis — Upload photo; AI suggests top 3 diseases and treatment protocols.
Open‑source datasets — Roboflow FishLens dataset contains over 15,000 labeled fish disease images.