Design And Implementation Of Automated Test Platform For Image Algorithm

As software technology is widely used in the Internet,people's quality requirements for software products are gradually increasing.So as to improve the quality of software,automated testing has become an important technology to reduce costs,shorten the development cycle,and ensure product quality.Due to the image algorithm is widely used in the field of short video and beauty camera,it greatly promotes the development and version iteration of the image algorithm,and the test workload is gradually increasing.Therefore,automated testing has become the main way to improve the efficiency of image algorithm testing.The automated test platform for image algorithms designed in this thesis combines actual business requirements.For each business requirement,the same points are extracted as common points,and different points are retained as customized points.Automatic testing is used instead of manual testing,which greatly saves testers from frequent construction.Test environment and time for manual operation of equipment.Mainly lies in the integrated automation test platform realized through the joint action of Jenkins continuous integration,business automation testing,equipment intelligent distribution and equipment scheduling,helping companies to quickly submit test scripts,reducing errors due to human factors,and greatly ensuring the quality of the program,To become an efficient,high-quality software automated testing platform.The automated test platform for image algorithms uses the SSM framework,Jenkins continuous integration,MySQL as the basic database,WebAssembly as the video parsing tool,FFmpeg for key frame extraction,and Python as the automated scripting language.This article is divided into seven modules,which are task management,automated scheduling,execution management,job management,equipment management,resource management,and result management.The task management module mainly includes task creation,viewing historical task list,etc.The automatic scheduling module is mainly based on scheduling algorithms to automatically schedule containers and equipment to perform tasks.Execution management is mainly to manage the execution information and process in the execution process.Job management mainly includes conFigureuring Job information and setting task execution steps,namely the pipeline.Equipment management mainly includes equipment access,equipment management,and equipment monitoring.Resource management mainly includes picture,video,magic expression,algorithm model package resource management.Result management is to analyze the result data of automated test runs and provide customized visual display on the front end,with functions such as logs,email notifications,image comparison,video comparison,calculation algorithm results,and test reports.At present,this platform has been put into use.The company's traditional method adopts a semi-automated test process,which requires humans to do some steps in person to make the test task go smoothly.The realization of automated testing does not require manual intervention,which helps save a lot of manpower and financial costs,and has been unanimously recognized by the company's personnel.At present,many businesses have begun to access the platform.