Electronic device and method for artificial intelligence detection

Abstract

The invention discloses an electronic device and method for realizing artificial intelligence detection. The device comprises a processor used for controlling components in the device; the backgroundcomponent is used for initializing an interactive interface after being started, initializing an artificial intelligence detection model and generating a plurality of artificial intelligence detectioninstances provided with mutual exclusion locks; when the interactive interface receives a calling request instruction from the foreground component, in response to the calling request instruction, the state of the mutual exclusion lock is obtained, and whether the artificial intelligence detection instance is currently available or not is judged; if so, the artificial intelligence detection instance is accessed through the interaction interface; and the artificial intelligence detection instance performs operation and feeds back an operation result to the foreground component through the interaction interface. The device can realize a sequential calling arrangement mechanism of the artificial intelligence detection instances, can ensure that the plurality of interactive interfaces orderlycall the artificial intelligence detection instances in sequence, and solves the problem of resource competition when the plurality of interactive interfaces call the same artificial intelligence detection instance at the same time.

Description

technical field [0001] The present application relates to the field of computer technology, in particular to an electronic device and method for realizing artificial intelligence detection. Background technique [0002] At present, AI (artificial intelligence) algorithms have been widely used in image recognition, speech recognition, text analysis, search engine ranking, recommendation system ranking and so on. With the emergence and popularity of various deep learning frameworks, more and more companies now choose to gradually apply and engineer their own AI algorithms. However, different deep learning frameworks use different methods. For example, Google’s open source TensorFlow comes with its own TFServer component, while PyTorch, MXNET and other frameworks are more for algorithm researchers to carry out research and development work, so that if you want to engineer some AI applications, you have to consider many compatibility factors. The most common application and eng...

AI Technical Summary

Problems solved by technology

[0003] However, AI algorithms are different from ordinary algorithms. Generally, AI algorithms rely on their own models and initialization configuration files. This makes each time an AI algorithm is run, there will be a long initialization phase for the AI ​​model and configuration. Loaded on the GPU computing device, this makes it necessary to re-execute the initialization work every time an HTTP request is received, thereby increasing the service delay of the HTTP interface, reducing work efficiency, and increasing the failure rate

Images