search

fyangneil / pavement-crack-detection

347 stars 77 forks source link

Plotting Precision-Recall Curve #20

Open ftlong6666 opened 3 years ago

ftlong6666 commented 3 years ago

Hello @fyangneil I have a question how to plot the precision and recall curve?, Can we use the mTpIUPR.txt that available in evalIu folder where the column indicates[threshold, IU, precision, recall]?

fyangneil commented 3 years ago

yes, you can

From: ftlong6666 notifications@github.com Sent: Wednesday, December 9, 2020 6:24 PM To: fyangneil/pavement-crack-detection pavement-crack-detection@noreply.github.com Cc: Fan Yang fyang@temple.edu; Mention mention@noreply.github.com Subject: [External] [fyangneil/pavement-crack-detection] Plotting Precision-Recall Curve (#20)

Hello @fyangneilhttps://github.com/fyangneil I have a question how to plot the precision and recall curve?, Can we use the mTpIUPR.txt that available in evalIu folder where the column indicates [threshold, IU, precision, recall]?

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHubhttps://github.com/fyangneil/pavement-crack-detection/issues/20, or unsubscribehttps://github.com/notifications/unsubscribe-auth/ALAUWTW4REI6IJ5WHG55QO3SUABJ3ANCNFSM4UUGN2XA.

ftlong6666 commented 3 years ago

Hi @fyangneil thank you for the reply. I have tried to use it. Basically I used your precomputed result. The number in ODS, OIS, and AIU are same. However when I tried to plot precision recall curve, I got this result which is quite different from your images in the paper. I used mTpIUPR.txt where the 4th column (Recall) as x-axis and 3rd column (precision) as y-axis. However I got this result which is quite different. Do you know why?, do you have a code to plot this? , should we sort the recall first before plotting?. Thank you so much for your help.

image

fyangneil commented 3 years ago

sorry, i cannot find the plot code anymore

From: ftlong6666 notifications@github.com Sent: Thursday, December 10, 2020 12:11 AM To: fyangneil/pavement-crack-detection pavement-crack-detection@noreply.github.com Cc: Fan Yang fyang@temple.edu; Mention mention@noreply.github.com Subject: [External] Re: [fyangneil/pavement-crack-detection] Plotting Precision-Recall Curve (#20)

Hi @fyangneilhttps://github.com/fyangneil thank you for the reply. I have tried to use it. Basically I used your precomputed result. The number in ODS, OIS, and AIU are same. However when I tried to plot precision recall curve, I got this result which is quite different from your images in the paper. I used mTpIUPR.txt where the 4th column (Recall) as x-axis and 3rd column (precision) as y-axis. However I got this result which is quite different. Do you know why?, do you have a code to plot this? , should we sort the recall first before plotting?. Thank you so much for your help.

[image]https://user-images.githubusercontent.com/56487515/101724203-f7d9c580-3ae8-11eb-96c7-815d9c17d14d.png

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHubhttps://github.com/fyangneil/pavement-crack-detection/issues/20#issuecomment-742244041, or unsubscribehttps://github.com/notifications/unsubscribe-auth/ALAUWTVICACBY7XNHMO56UDSUBKAVANCNFSM4UUGN2XA.

ftlong6666 commented 3 years ago

@fyangneil In the measurement do you use any transitional areas like thinning the edges? I see in this script crack_nms.m where we have this command predmap = edgesNmsMex(predmap, O, 2, 5, 1.01, 8);. So what is the meaning of 2 and 5 here?, do you use 5 pixel for thinning the edges?

fyangneil commented 3 years ago

you can find the definition of the function and its parameters https://github.com/fyangneil/pavement-crack-detection/blob/master/eval_tool/edges/privated/edgesNmsMex.cpp

From: ftlong6666 notifications@github.com Sent: Friday, December 11, 2020 2:21 PM To: fyangneil/pavement-crack-detection pavement-crack-detection@noreply.github.com Cc: Fan Yang fyang@temple.edu; Mention mention@noreply.github.com Subject: [External] Re: [fyangneil/pavement-crack-detection] Plotting Precision-Recall Curve (#20)

@fyangneilhttps://github.com/fyangneil In the measurement do you use any transitional areas like thinning the edges? I see in this script crack_nms.m where we have this command predmap = edgesNmsMex(predmap, O, 2, 5, 1.01, 8);. So what is the meaning of 2 and 5 here?, do you use 5 pixel for thinning the edges?

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHubhttps://github.com/fyangneil/pavement-crack-detection/issues/20#issuecomment-743380589, or unsubscribehttps://github.com/notifications/unsubscribe-auth/ALAUWTQ6A6GXNLTLTXA2CXLSUJWLFANCNFSM4UUGN2XA.