search

yujheli / Pitt-Radar

The repository is the release dataset of our Pitt-Radar of the CVPR 2023 paper "Azimuth Super-Resolution for FMCW Radar in Autonomous Driving"
MIT License
22 stars 2 forks source link

Questions about the super-resolution performance #8

Closed RoboticsZhang closed 8 months ago

RoboticsZhang commented 8 months ago

Hello, thank you for your novel and solid work, as well as the code and data open-sourcing. According to the paper and the code, it seems that the ADC-SR takes adc data from 4 receivers as the input, and that from 12 receivers as the output. I wonder that if ADC-SR can take adc data from 12 receivers as the input, and output adc data from more hallucinated receivers? That is, given a single-chip radar with 3TX and 4RX, could we use ADC-SR to generate super -resolution adc data with more hallucinated receivers? Thank you!

yujheli commented 8 months ago

We have tried and that also works for generating super-super resolution. However, since we do not have the real high number of receivers, there is no way to verify the higher settings.

RoboticsZhang commented 8 months ago

We have tried and that also works for generating super-super resolution. However, since we do not have the real high number of receivers, there is no way to verify the higher settings.

Thank you for your reply. I will close this issue then.

yujheli commented 7 months ago

I found the visualization of super-super resolution I had last year below (4 receivers vs 12 receivers vs 108 receivers):

image
RoboticsZhang commented 7 months ago

Thank you!