search

brade31919 / radar_depth

Source code of the IROS 2020 paper "Depth Estimation from Monocular Images and Sparse Radar Data"
MIT License
87 stars 13 forks source link

Evaluation of Pretrained Models #18

Open Abdulaaty opened 2 years ago

Abdulaaty commented 2 years ago

Hello, I tried to evaluate the models given, I also followed the suggestion provided in #11

But then there are errors generated due to the model creation in line 207

https://github.com/brade31919/radar_depth/blob/5e6e75772ff379aac65379a50d4042a7c64c869d/main.py#L207

In order to get the code running, I commented some of the conditions that address the args.evaluate in the method: https://github.com/brade31919/radar_depth/blob/5e6e75772ff379aac65379a50d4042a7c64c869d/main.py#L64

Although some inconsistences are available, but I got the code to evaluate the available pretrained models in: https://github.com/brade31919/radar_depth/blob/5e6e75772ff379aac65379a50d4042a7c64c869d/README.md#downlaod-the-pre-trained-models

The results are really different from the paper (RMSE is in magnitude of 20). Is there something wrong, or there're a problem with the available pretrained models?