Closed lfrcoder closed 6 years ago
We didn't follow the original paper preprocesses or data augmentation. The basic configuration used in the experiments for this dataset can be found in /code/config/tt100k_[detection or classification].py, the last two blocks of this file show if we used any preprocessing at all or data aug. Think this was a Master Lab session, we didn't try to replicate papers, only try to make things work and play with hyperparameters.
Regards.
Thank you for your answer. I tried the TT100K experiment on YOLO-V3, but the detection was not good. wonder if it is because the pre-processing is different from the original author.
I would like to ask some questions: Have you done the same preprocessing as original author to preprocess the data for the TT100K database?Such as perspective transformation, as if not mentioned in your experiment.