Garrick Brazil

I have narrowed the issue down to this row of my spawnpoints table, unfortunately I do not see anything wrong with it. For additional context, this is row 662 out...

The below data should be downloaded and extracted - the left color images - camera calibration matrices - training labels to be more clear the "data/kitti" folder should look something...

We have not tried Resnet in a long time. If I recall the performance is slightly less than with DenseNet. I can plan to add a few more backbone models...

I do not have time currently to test the models. But you can feel free to adjust the stride of ResNet backbones such that the final stride is still 16....

I've attached examples for each ResNet18 and ResNet50 [here](https://github.com/garrickbrazil/M3D-RPN/files/3768412/resnet_models.zip). However, I have NOT tested them :)

Attached are fixed model files (I briefly check that they begin training on my machine). [resnet_models_fixed.tar.gz](https://github.com/garrickbrazil/M3D-RPN/files/3774966/resnet_models_fixed.tar.gz) Regarding the performance you posted, that is very alarming. If the above models converge...

Can you provide details on your python version and packages, etc? It seems that for this type of error there is probably something inconsistent in that realm. You can refer...

Hi Haorui-Ji, we use the anchors generated at training for inference at test time! Hence, after the anchors have been determined at training time, they are fixed

My feeling is that this is the train -> train accuracy, which looking at the results suggests a bit of over fitting compared to the validation. Nevertheless, I am guessing...

Would it be possible to provide an example of how to use the SUN RGB-D model for (v3 | [email protected]: 43.7 which uses 20211007_105247.pth and imvoxelnet_total_sunrgbd_fast.py) model, with an arbitrary...