about data aug

[hello7623]

Do you have any code for data augmentation，I'd like to study, could I ？

[skokec]

You can use the attached code (DataAugmentator.tar.gz), however, this code is in matlab and is based on our raw dataset images before we exported it for the public, so you can use it mostly for looking-up the details. The usage of this code and the entrypoints are below:

aug = DataAugmentator();                    
aug.collectData(srcDetailedAnnoNames, srcAnnoNames, srcImgNames);                    
aug.prepareDataDistributions(templates);

generatedSamples = aug.generateFromRealImages(selectedClasses_forGenericImgs_Fr, numTrain_instances_selectedClasses_forGenericImgs_Fr, self.outDirAnnot, self.outDirImgs, self.outDirExtractedTS, imgCounterAbsolute);

Hope this helps in any way.

Best, Domen

[hello7623]

Thanks！

------------------ 原始邮件 ------------------ 发件人: "Domen Tabernik"<notifications@github.com>; 发送时间: 2020年4月28日(星期二) 晚上9:22 收件人: "skokec/detectron-traffic-signs"<detectron-traffic-signs@noreply.github.com>; 抄送: "Faker"<981839079@qq.com>;"Author"<author@noreply.github.com>; 主题: Re: [skokec/detectron-traffic-signs] about data aug (#5)

You can use the attached code (DataAugmentator.tar.gz), however, this code is in matlab and is based on our raw dataset images before we exported it for the public, so you can use it mostly for looking-up the details. The usage of this code and the entrypoints are below: aug = DataAugmentator(); aug.collectData(srcDetailedAnnoNames, srcAnnoNames, srcImgNames); aug.prepareDataDistributions(templates); generatedSamples = aug.generateFromRealImages(selectedClasses_forGenericImgs_Fr, numTrain_instances_selectedClasses_forGenericImgs_Fr, self.outDirAnnot, self.outDirImgs, self.outDirExtractedTS, imgCounterAbsolute);

Hope this helps in any way.

Best, Domen

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or unsubscribe.