Sebastian Rietsch

I trained my detector on a specific dataset where it is performing pretty well. Now I want to finetune the model to another image domain, object classes stay the same....

[Enhancement] Multiple model iterations per Optuna trial and mean performance objective

5

I currently have the problem that, a lot of times, the results Optuna optimization produces are not really too optimal, due to the stochastic nature of RL training. For example,...

SUMO integration

1

Could somebody explain to me how the SUMO integration has been implemented? Does the esmini player start a SUMO server somewhere? Or is there a way to interact with the...

Zotero note formatting not correctly imported, Comments not imported

4

Hi, first of all: Thank you so much for this, bibnotes is awesome! Two issues I encounter when I use your Addon is that 1. highlight comments are not imported,...

Hi, FYI, I created a PyTorch implementation of ForkGAN (https://github.com/seawee1/ForkGAN-pytorch.git). It's based on the CycleGAN repository and features some neat little extras (MultiGPU training, instance-level losses, automatic mixed precision for...

Sphinx documentation

2

Besides code documentation, we would like to provide a [Sphinx](https://www.sphinx-doc.org/en/master/) for Driver Dojo.

More efficient add_batch_fn implementation

1

Really much appreciate this repository. However, for non-trivial batch sizes, the `add_batch_fn` caused a lot of slowdowns due to having to add every batch element sequentially. In case somebody is...