Open FaisalAhmed-NSL opened 1 year ago
Dear : I m sorry to return a late reply. About Q1 and Q2, I am afraid i can not help since its realized by other robot companies. About Q3, i believe the only thing you need to ensure is about the quality of captured images, since blurred and broken images will influence the recognition results. About Q4, the data preprocessing steps are simple, which can be seen in util/augmentation.py, we only use size-transformation related operation, but some other augmentation methods such as RandomBrightness, RandomContrast, RandomMirror can also be applied.
Best wishes yan
Thank you very much for your useful information.
Hello Shuyansy! Can you please give some information about the collection of datasets?I have some queries for you. I'm interested to learn more about this project.