Issues of evaluate code

[heng94]

Hi ,
when I run the evaluate.py , There is only " gt" result , no "prediction " result .What code should I modify to get the "prediction" result ! Thanks

[bertjiazheng]

I have the same issue, is anyone able to solve it?

[DRAhmadFaraz]

@HanochZzhou @bertjiazheng Can you show your gt" results screenshot.?
let me help you then.

[bertjiazheng]

Hi, @DRAhmadFaraz, I have got the predictions of PlaneNet, but the per-pixel recall seems to be lower than that in the paper. Can you reproduce the evaluation in the paper?

Here are the results
Per-Pixel Recall: [0. , 0.11259292, 0.39307635, 0.50202066, 0.53868158, 0.55395703, 0.57877339, 0.57877339, 0.60665465, 0.60665465, 0.60665465, 0.62022231, 0.62022231]
Per-Plane Recall: [0. , 0.16, 0.32, 0.36, 0.38, 0.42, 0.46, 0.46, 0.48, 0.48, 0.48, 0.5 , 0.5 ]

[jcliu0428]

I have the same issue.Anyone knows how to fix it?

[jcliu0428]

@bertjiazheng Have you got the prediction result in your 'evaluate folder'? Did you fix your command line to run evaluate.py? Thanks.

[bertjiazheng]

@jcliu0428

Run with command line:

python evaluate.py --dataFolder=*** --numImages=760 --methods=2000000 --numOutputPlanes=10

[jcliu0428]

@bertjiazheng Have you change anything in checkpoint.ckpt?
When I try your command line,it shows that:
InvalidArgumentError (see above for traceback): Unsuccessful TensorSliceReader constructor: Failed to get matching files on checkpoint/planenet_hybrid3_bl0_dl0_ll1_pb_pp_ps_sm0/checkpoint.ckpt: Not found: checkpoint/planenet_hybrid3_bl0_dl0_ll1_pb_pp_ps_sm0; No such file or directory

[bertjiazheng]

@jcliu0428 You don't need to modify anything. Make sure the methods=2000000, 0 will disable the corresponding method.

[jcliu0428]

@bertjiazheng I modify ALL_METHODS,it also works.Thanks a lot!

[XYZ-qiyh]

Hello, I'm confused about the python evaluate.py output result

what does this mean?
Emm, where should I find the recall ?
@bertjiazheng @art-programmer Thanks a lot~

[bertjiazheng]

@QTODD you can find the pixel recall here and plane recall here.

[XYZ-qiyh]

Hey, when I run the evaluate.py, I only got the prediction of PlaneNet.
What can I do to get the results of other existing methods ?? @jcliu0428 @bertjiazheng

[skq-5233]

Why is there no quantitative comparison output when I run evaluate.py?Looking forward to your reply. Thank you!@bertjiazheng

[skq-5233]

Hi, @DRAhmadFaraz, I have got the predictions of PlaneNet, but the per-pixel recall seems to be lower than that in the paper. Can you reproduce the evaluation in the paper?

Here are the results
Per-Pixel Recall: [0. , 0.11259292, 0.39307635, 0.50202066, 0.53868158, 0.55395703, 0.57877339, 0.57877339, 0.60665465, 0.60665465, 0.60665465, 0.62022231, 0.62022231]
Per-Plane Recall: [0. , 0.16, 0.32, 0.36, 0.38, 0.42, 0.46, 0.46, 0.48, 0.48, 0.48, 0.5 , 0.5 ]

Hello, when I run python evaluate.py, I do not know what the reason is and do not output quantitative results such as rel, rmse, etc. I hope you take the time to reply to me. Thank you!