-
Notifications
You must be signed in to change notification settings - Fork 213
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Obtaining per-point predictions for original point cloud #159
Comments
Hi, you can access the confusion matrix after evaluation with
|
Hi, Thanks for the quick response. Yes, this is what I'm currently using. My question is, is it possible to obtain the prediction for every individual point, not the accumulated confusion matrix. The reason why I'm asking is because I want to visualize the results, I want to see where the errors were made, and I can't do it using just the confusion matrix. Is this possible? |
You can use the function |
Will check it, will certainly be better than only a confusion matrix. Just to clarify, there is then no way to actually export the per point predictions? Visualizing is nice, but I can't really compare it to the results of another algorithm and compare what each method is better at. For that, I'd need to import the predictions into my own tool for visualization. |
Hi, you can access to the output of the network through If you want the result on the original point cloud you can upsample the prediction |
Thank you, this is exactly what I needed. Will let you know how it goes. |
of note, I highly do not recommend to write the output in a csv file, as it is very slow compared to a h5 file for example. this is only done to comply with semantic3d's submission requirements. |
Noted |
Hi, I think I've gotten stuck a bit. There's 2 places where there's an |
Actually I see in the way the confusion matrix is computed, that these are actually all of the points but partitioned into 64 batches, and each point in the batch is assigned the same label. Is this an error? I'm a bit confused. |
Hello,
I've used your method with a custom dataset, posed basically as a binary classification problem. I am able to obtain the confusion matrix, but not an actual array of predictions for each point in the original pointcloud. Is this possible?
The text was updated successfully, but these errors were encountered: