FISHERIES RESEARCH INSTITUTE,MOA,TAIWAN

The control of bait feed is an important link for aquaculture and fisheries. In order to assist net cage aquaculture farmers in feeding management, the Fisheries Research Institute, Council of Agriculture has researched and developed a “net cage aquaculture creature activity monitoring system”. The aim is to judge whether the fishes get full according to the degree of splashing and movement state during the process of fighting for food after feed input with the help of underwater image analysis. Thus, excessive feed input can be avoided, and the activity situation and health condition of fishes can be observed.

The institute has pointed out that the expenditure of bait feed is about 50% of the cost of net cage aquaculture. If the waste can be reduced, the aquaculture cost can be greatly reduced, the profits can therefore rise, and pollution problems can decrease as well, so as to keep the aquaculture environment clean and the fishes healthy. At present, the main fingerlings of maritime net cage aquaculture industry in Taiwan are trachinotus blochii, grouper, and cobia. Manual feeding usually requires the skillful experience of a worker to judge whether the feeds are sufficient and whether the fishes get full, or whether there is any abnormal condition. As specific quantized data are lacking for reference, it is easy to confront a situation of excessive feeding with the purpose of avoiding fishes from not getting full.

The “net cage aquaculture creature activity monitoring system” developed by the Fisheries Research Institute sets up waterproof cameras on the water and under the water. Visual analysis software is equipped for the operation host, and “splash sensing algorithm” and “fish movement quantization algorithm” are applied for calculation, while the environment on the water is analyzed by “splash sensing algorithm”. Generally, in the early period of feeding, the degree of splashing is usually great as fishes fight for feeds; and in the later period, there is usually no splash. The system uses visual software to separate splashing from background and judges the degree of splashing from the scope and size of splash distribution as the basis of whether the fishes get full. The underwater environment applies “fish movement quantization algorithm” to analyze the activity of fishes by the disturbance amplitude of the pixel value.

The Fisheries Research Institute stated that the system is now being tested in private trachinotus blochii net cage culturing farms, with the aim to assist farmers at determining the bait feed replacement and unit amount input and to provide a precise reference for feeding management. It is estimated that if 10% expenditure of feed can be reduced, then 5% production cost can be lowered. In other words, every net cage can save NT$200,000 in feed expenditure, thus greatly improving aquaculture and production effectiveness.