This technology uses a customized high-resolution saltwater proof video camera that operates underwater for months on end without maintenance. The captured images are pre-processed to adjust the brightness and maximize the sharpness, after which the images are input into a region-based convolutional neural networks (R-CNN) for training and identification. This technology is currently able to reliable identify shrimp in these images and to distinguish between the shrimp’s head and tail in order to conclude whether the image shows a complete shrimp.

Like other sectors in the global aquaculture industry, shrimp farming depends on human judgement and experience and requires observation of the conditions in the pond as well as removal of the shrimp from the water to conduct observations and measurement of the body length. In order to allow the breeder to observe the growth of the shrimp more efficiently in their natural underwater environment, this system uses a seawater channel to corral the shrimp in front of an underwater camera, which then sends a stream of images to a host computer for shrimp identification. Shrimps are identified by the presence of both a head and body part, and the weight of the shrimp is obtained by using a formula that converts length to weight. 

By developing an artificially intelligent (AI) system that recognizes shrimp images and automatically measures the shrimp’s length and weight, this technology is intended to free shrimp breeders from traditional methods (i.e. human eyes and experience) of observing growth in shrimps. This system will not only reduce labor-force requirements in the future, but also allow for more immediate observation of larger quantities of shrimp. This in turn will allow farmers to quickly adjust the feed supply to optimize the growth rate of the shrimp.

如對技術有興趣，歡迎與我聯繫，謝謝!