Mouse activeness detection method and system based on deep learning and sanitation evaluation method

Abstract

The invention discloses a mouse activity detection method and a mouse activity detection system based on deep learning, and a sanitation evaluation method. The method comprises the steps: obtaining amouse activity video, and obtaining a training data set; constructing a RetinNet network model, and importing the training data set into the RetinNet network model for training to obtain a mouse detection model; utilizing ImageAI to call a mouse detection model to identify a target object in each frame of picture of the to-be-detected video, and outputting a candidate box of the target object in each frame of picture and a corresponding class probability; according to the candidate boxes and the class probabilities, filtering the candidate box corresponding to each frame of picture by adoptinga non-maximum suppression algorithm to obtain a target box set for each frame of picture; and according to each frame of picture and the target frame set, adopting a target counting algorithm to obtain the occurrence frequency of mice in the to-be-detected video, and evaluating whether the environmental sanitation is qualified or not according to the occurrence frequency. According to the method,a large amount of manpower is saved, the effectiveness and reliability of target object detection are high, and good reference data can be provided for sanitation evaluation.

Description

technical field [0001] The application belongs to the field of target detection, and specifically relates to a mouse activity detection method and system based on deep learning, and a sanitation evaluation method. Background technique [0002] Rats are a species with strong reproductive ability, and rats can spread many kinds of diseases, which seriously endanger human health. Rats often haunt sewers, toilets, kitchens, debris piles, garbage dumps, etc., move back and forth between bacteria-carrying places and clean places, and spread pathogenic bacteria through rat feet, body hair and stomach contents. According to statistics, there are more than 50 kinds of diseases transmitted by mice, such as plague, epidemic hemorrhagic fever, typhus, etc. These diseases are not only highly contagious, but even life-threatening in severe cases. [0003] The kitchen sanitation of catering companies has always been concerned, and rat traces are a relatively intuitive reflection of the ki...

AI Technical Summary

Problems solved by technology

[0003] The kitchen sanitation of catering companies has always been concerned, and rat traces are a relatively intuitive reflection of the kitchen sanitation. In the prior art, the statistics of rat holes, rat feces, rat bite marks or the use of sticky mouse boards are usually used. The splint method, stealing method, visual inspection, etc. are used to count mice, but these methods use a lot of manpower for statistics, the process is cumbersome, and the statistical error is often large; or the infrared sensor is used for detection, but this method cannot distinguish whether it is the target object. Easy to false detection

Images