Sudo42b comments

Results 8 comments of


                                            Sudo42b

More RAM is consumed as training epoch increases

reduce num_worker thread and batch size.

dataset structure

Please refer to [official website](https://www.ei.tum.de/mmk/dad/) or [Gdrive](https://drive.google.com/drive/folders/1TVzNODlKiRKj60cwadnXofEsBkCDl1DZ?usp=sharing).

how to integrate with detectron2?

@jinfagang This **[Link](https://github.com/MarkPotanin/copy_paste_aug_detectron2.git)** @

Exception 'out_of_range' not member of std

Thx!

reward inialiser value

Please refer this [link](https://github.com/aitorzip/DeepGTAV/issues/118#issue-568903232) page add `#include ` in `LaneRewarder.cpp`

Unable to download vww pretrained models

> I am getting a timeout error when downloading https://hanlab.mit.edu/projects/tinyml/mcunet/release/mcunet-10fps_vww.json. Is the link currently down? I am not the administrator of this repository, but you can use the following url:...

Inconsistency of CMSIS-NN Quantization Method(Q-format) with ARM Documentation

Hello @mansnils , thank you for your response. Unfortunately, it seems that my limited fluency in English may have caused some misunderstandings, for which I apologize for any confusion caused....

Inconsistency of CMSIS-NN Quantization Method(Q-format) with ARM Documentation

Dear @mansnils, Thank you for your response. Upon reviewing the CMSIS-NN documentation, I confirmed that the presence of _s indicates compatibility with TensorFlow Lite Micro. Additionally, when examining the algorithm,...