How to create my own dataset?

[chan-yuu]

Nice job. I want to know what tools you used to create the dataset, is it open source? Mainly the content of the language text that needs to be marked in the driving process.

[ChonghaoSima]

Hello, we work with an industrial annotation team: https://www.molardata.com/ . They provide annotations from human expert drivier on DriveLM selected key frames. I think they didn't open-source the labelling tool and planform, but the functionalities are mostly implemented in https://github.com/wkentaro/labelme and similar open-sourced annotation planform.

[chan-yuu]

Thanks! I see.