Learning from Data (Fall 2022)

[kundtx]

http://8.129.175.102/lfd2022fall-poster-session/13.html

[charon-bo]

Group22 Shilong Mu:Impressive work , in your experimental section "LIEVO RMSE Accuracy", you use the TUM dataset for testing, but as far as I know, the TUM dataset does not have test sequences under low light conditions. How did you do the testing?

[qingyuhai]

@charon-bo Group22 Shilong Mu:Impressive work , in your experimental section "LIEVO RMSE Accuracy", you use the TUM dataset for testing, but as far as I know, the TUM dataset does not have test sequences under low light conditions. How did you do the testing?

G13, Qian Zhang: Hi Shilong, thank you very much for your question. You have asked a very professional question.Indeed, as you know, the current TUM dataset only has sequences for normal light, not for low light. So during the experimental test, we try to adjust the TUM data brightness to make the dataset less bright. Specifically, first, we choose to convert the RGB data into RAW data by unprocessing the original normal light data. Secondly, the RAW data is linearly transformed, so that the brightness of the image is weakened to become dark. Finally, the RAW data is then converted back to RGB data for testing. Thanks again for your question.