David María Arribas

Thank you!!! I've found it very usefull, the reprojected images and disparities in tensorboard look very promising, considering training was done with only 8k images... now I'm training with more...

Hi!!!! I have training with 31k images, and got some results. If someone is interested, here [TensorBoard](http://dmariaa.es:6006) is the tensorboard obtained, and here [ETS2 Dataset](https://www.dmariaa.es:8101/files) the dataset can be reviewed...

> @dmariaa I know it's been late to response, but your idea of using Euro Track Simulator to generate dataset is quite interesting. > For the first issue, complete black...

Absolutly Yes @Apoorvgarg-creator .... I would consider adding a field to the exercise model in the Django backend, something like "tags"... then you can use that field in the backend...

BTW... Tags have to be maintained (added, modified, removed) in the /admin section of robotics academy, the list frontend need to have the possibility of being filtered by this tag,...

Linked a new branch, "new-architecture" to work on this issue... More info about it in this discussion #1799

Any news about this?