Less-sample sluice image classification method

The invention discloses a few-sample sluice image classification method, which combines double attention and time sequence convolution to establish a meta-learning model, arranges and arranges a data set according to a small sample learning standard, and further performs feature extraction, meta-learning training and operation state recognition on the data set. The sluice opening and closing states can be accurately and automatically classified across sluice types under the condition of few samples. A memory enhancement model is formed through interlaced combination of double attention and time sequence convolution, useful general experience in knowledge can be learned and memorized in a high-bandwidth mode, new knowledge can be learned rapidly through the experience, and high opening and closing state classification accuracy can be obtained across sluice types. The method gets rid of the dependence of an existing water conservancy field deep learning algorithm on a large amount of training data and the limit