search

BSI-OFIQ / OFIQ-Project

Open Source Facial Image Quality
Other
38 stars 12 forks source link

Not sure how to decide which image is good #16

Closed tadam98s closed 3 months ago

tadam98s commented 5 months ago

table.xlsx 2021-MickeyCohen All the images habe UQS ~ 27. Please include the "perfect image" in the test set.

Also, you are using the German convention of ";" while the default elsewhere is "," Comma Separated Values. Can you add a parameter to set it. You can keep the ";" as default.

It looks good. I will study it in depth. Thank you for adding the "all images" option.

Best, Mickey

bbtams commented 5 months ago

Hi Mickey,

I suppose you only looked at the UnifiedQualityScore column which is the native quality measure; you should inspect the colum UnifiedQualityScore.scalar where you can find the quality component measure between 0 (worst) and 100 (best). I don't think there is such a thing as the "perfect image" among real ones.

We used the ";" to enable loading of the CSV file with Excel; I am not sure if this only because this is German convention. I can create a ticket for integrating configuration of the CSV separation character, but this will have low priority as it only affects the sample application (and not the OFIQ library).

Thank your for studiying OFIQ in depth; any improvement suggestion would be helpful.

Best, Benjamin

tadam98s commented 5 months ago 
body p { margin-bottom: 0cm; margin-top: 0pt; } 

Hi,

Same question which threshold indicates "take
    the face it meets 17974-5 requirements".

B-01 with smile gets 65 - smile is not
    permitted
B-02 rolled gets 65 -
    strong roll is not allowed
B-09 Background gets 67 - background is not
    allowed
C-01 gets 77 but head is too small compared
    with the standard (Which parameter gives the eye distance? Is it
    >> 90?)
C-06 glasses get 53 - occlude the eyes I think
    it should be lower

C-07 two faces gets 82 - but two faces not
    allowed
C-10 Background gets 86 - background is not
    allowed and also face is too small. (Which parameter gives eye
    distance in pixels ?)
R-02 rolled gets 71 - strong roll is not
    allowed
R-04 eyes closed head covered 78 - both not
    allowed
R-07 scarf gets 56 - chin is not visible I
    would prefer a lower score.

R-09 Background gets 83 - background is not
    allowed and also face is too small. (Which parameter gives eye
    distance in pixels ?)

  body p { margin-bottom: 0cm; margin-top: 0pt; } 

  Best regards,
   Mickey Cohen
   Shanit Ltd.
   CEO
   POB 23410, Jerusalem 9123302, Israel
    Schedule a meeting
    (M) +972-54-758-6312
   Skype: tadam_98 
   ***@***.***

On 21/03/2024 9:23 AM, bbtams wrote:

  Hi Mickey,
  I suppose you only looked at the UnifiedQualityScore
    column which is the native quality measure; you should inspect
    the colum UnifiedQualityScore.scalar where you can find the
    quality component measure between 0 (worst) and 100 (best). I
    don't think there is such a thing as the "perfect image" among
    real ones.
  We used the ";" to enable loading of the CSV file
    with Excel; I am not sure if this only because this is German
    convention. I can create a ticket for integrating configuration
    of the CSV separation character, but this will have low priority
    as it only affects the sample application (and not the OFIQ
    library).
  Thank your for studiying OFIQ in depth; any
    improvement suggestion would be helpful.
  Best,
    Benjamin
  —
    Reply to this email directly, view it on GitHub, or unsubscribe.
    You are receiving this because you authored the thread.Message
      ID: ***@***.***>
  [

{ @.": "http://schema.org", @.": "EmailMessage", "potentialAction": { @.": "ViewAction", "target": "https://github.com/BSI-OFIQ/OFIQ-Project/issues/16#issuecomment-2011366512", "url": "https://github.com/BSI-OFIQ/OFIQ-Project/issues/16#issuecomment-2011366512", "name": "View Issue" }, "description": "View this Issue on GitHub", "publisher": { @.": "Organization", "name": "GitHub", "url": "https://github.com" } } ]

--------------1jKwxgYezhBCiNfc2WH8HUwI Content-Type: image/jpeg; name="0eW82mw1VPEtnBp9.jpg" Content-Disposition: inline; filename="0eW82mw1VPEtnBp9.jpg" Content-Id: Content-Transfer-Encoding: base64

/9j/4AAQSkZJRgABAQEAYABgAAD/2wBDAAIBAQIBAQICAgICAgICAwUDAwMDAwYEBAMFBwYH BwcGBwcICQsJCAgKCAcHCg0KCgsMDAwMBwkODw0MDgsMDAz/2wBDAQICAgMDAwYDAwYMCAcI DAwMDAwMDAwMDAwMDAwMDAwMDAwMDAwMDAwMDAwMDAwMDAwMDAwMDAwMDAwMDAwMDAz/wAAR CAAMAAwDASIAAhEBAxEB/8QAHwAAAQUBAQEBAQEAAAAAAAAAAAECAwQFBgcICQoL/8QAtRAA AgEDAwIEAwUFBAQAAAF9AQIDAAQRBRIhMUEGE1FhByJxFDKBkaEII0KxwRVS0fAkM2JyggkK FhcYGRolJicoKSo0NTY3ODk6Q0RFRkdISUpTVFVWV1hZWmNkZWZnaGlqc3R1dnd4eXqDhIWG h4iJipKTlJWWl5iZmqKjpKWmp6ipqrKztLW2t7i5usLDxMXGx8jJytLT1NXW19jZ2uHi4+Tl 5ufo6erx8vP09fb3+Pn6/8QAHwEAAwEBAQEBAQEBAQAAAAAAAAECAwQFBgcICQoL/8QAtREA AgECBAQDBAcFBAQAAQJ3AAECAxEEBSExBhJBUQdhcRMiMoEIFEKRobHBCSMzUvAVYnLRChYk NOEl8RcYGRomJygpKjU2Nzg5OkNERUZHSElKU1RVVldYWVpjZGVmZ2hpanN0dXZ3eHl6goOE hYaHiImKkpOUlZaXmJmaoqOkpaanqKmqsrO0tba3uLm6wsPExcbHyMnK0tPU1dbX2Nna4uPk 5ebn6Onq8vP09fb3+Pn6/9oADAMBAAIRAxEAPwD79PxatfG3wvj1jSxqOpeJLPUJE1tUYzP5 UpBgmEYy20sShYA4f72NyVgaX+1xqvhSOWzs9UvtPVJW82FZWj2SDhgV7EYwQeeK+afi5daj +zN+0/4lsvB+taxpcnhjVJ7awu1uSLhEDYwzAAMMcEEYIHIPNfU37Gn7Jeg/tvfDTU/iF8Qt W8Sal4o1PWJobm5guo7dZBHHEF+RY9o444AHA4r+NsDXxGPxX1fBy9niYe7J3tCSjo5cy97m k7XTi7u8m9bH8K5dxBmGZ4v6ngpezxdO8ZO7UJKGjkpJc/M3a6cWm7yck3Y//9k= --------------1jKwxgYezhBCiNfc2WH8HUwI Content-Type: image/jpeg; name="pbWUC04vBQSqgE1h.jpg" Content-Disposition: inline; filename="pbWUC04vBQSqgE1h.jpg" Content-Id: Content-Transfer-Encoding: base64

/9j/4AAQSkZJRgABAQEAYABgAAD/2wBDAAIBAQIBAQICAgICAgICAwUDAwMDAwYEBAMFBwYH BwcGBwcICQsJCAgKCAcHCg0KCgsMDAwMBwkODw0MDgsMDAz/2wBDAQICAgMDAwYDAwYMCAcI DAwMDAwMDAwMDAwMDAwMDAwMDAwMDAwMDAwMDAwMDAwMDAwMDAwMDAwMDAwMDAwMDAz/wAAR CAAMAAwDASIAAhEBAxEB/8QAHwAAAQUBAQEBAQEAAAAAAAAAAAECAwQFBgcICQoL/8QAtRAA AgEDAwIEAwUFBAQAAAF9AQIDAAQRBRIhMUEGE1FhByJxFDKBkaEII0KxwRVS0fAkM2JyggkK FhcYGRolJicoKSo0NTY3ODk6Q0RFRkdISUpTVFVWV1hZWmNkZWZnaGlqc3R1dnd4eXqDhIWG h4iJipKTlJWWl5iZmqKjpKWmp6ipqrKztLW2t7i5usLDxMXGx8jJytLT1NXW19jZ2uHi4+Tl 5ufo6erx8vP09fb3+Pn6/8QAHwEAAwEBAQEBAQEBAQAAAAAAAAECAwQFBgcICQoL/8QAtREA AgECBAQDBAcFBAQAAQJ3AAECAxEEBSExBhJBUQdhcRMiMoEIFEKRobHBCSMzUvAVYnLRChYk NOEl8RcYGRomJygpKjU2Nzg5OkNERUZHSElKU1RVVldYWVpjZGVmZ2hpanN0dXZ3eHl6goOE hYaHiImKkpOUlZaXmJmaoqOkpaanqKmqsrO0tba3uLm6wsPExcbHyMnK0tPU1dbX2Nna4uPk 5ebn6Onq8vP09fb3+Pn6/9oADAMBAAIRAxEAPwD7S+Fn7YPh/wCKVt8QLe6gjnuvAfgTWPF0 0i3TBp3smjwpVDny9sqhsDcWBIYZAr55T/go6/iK+vLjRF1nTdPMiqsCyNtVxGm8gkglS24j PIGASSM15PBrmqfs6/Enx9Y+E9QWxh1d77w/qDXGnWV619YNMyyW7+fC48uQKvmIoCvtXcDt XHo37Hf7J/hv4m/DTUL6+utYhlg1SS2VbeWMKVEMLZO6NjnLHv6V/Qb4NwuGjKvU1i7cuuvn dbLXazfyPyZcRV60o0oaSV7/AKWe/wCB/9k= --------------1jKwxgYezhBCiNfc2WH8HUwI Content-Type: image/jpeg; name="LUcLHC81mrGbmggP.jpg" Content-Disposition: inline; filename="LUcLHC81mrGbmggP.jpg" Content-Id: Content-Transfer-Encoding: base64

/9j/4AAQSkZJRgABAQEAYABgAAD/2wBDAAIBAQIBAQICAgICAgICAwUDAwMDAwYEBAMFBwYH BwcGBwcICQsJCAgKCAcHCg0KCgsMDAwMBwkODw0MDgsMDAz/2wBDAQICAgMDAwYDAwYMCAcI DAwMDAwMDAwMDAwMDAwMDAwMDAwMDAwMDAwMDAwMDAwMDAwMDAwMDAwMDAwMDAwMDAz/wAAR CAAMAAwDASIAAhEBAxEB/8QAHwAAAQUBAQEBAQEAAAAAAAAAAAECAwQFBgcICQoL/8QAtRAA AgEDAwIEAwUFBAQAAAF9AQIDAAQRBRIhMUEGE1FhByJxFDKBkaEII0KxwRVS0fAkM2JyggkK FhcYGRolJicoKSo0NTY3ODk6Q0RFRkdISUpTVFVWV1hZWmNkZWZnaGlqc3R1dnd4eXqDhIWG h4iJipKTlJWWl5iZmqKjpKWmp6ipqrKztLW2t7i5usLDxMXGx8jJytLT1NXW19jZ2uHi4+Tl 5ufo6erx8vP09fb3+Pn6/8QAHwEAAwEBAQEBAQEBAQAAAAAAAAECAwQFBgcICQoL/8QAtREA AgECBAQDBAcFBAQAAQJ3AAECAxEEBSExBhJBUQdhcRMiMoEIFEKRobHBCSMzUvAVYnLRChYk NOEl8RcYGRomJygpKjU2Nzg5OkNERUZHSElKU1RVVldYWVpjZGVmZ2hpanN0dXZ3eHl6goOE hYaHiImKkpOUlZaXmJmaoqOkpaanqKmqsrO0tba3uLm6wsPExcbHyMnK0tPU1dbX2Nna4uPk 5ebn6Onq8vP09fb3+Pn6/9oADAMBAAIRAxEAPwD78/Y9/Y/0fxv4Q1jxP4o1jwv4iuPEOhy6 hDDf3H+maVc5z9onySdmWJLMcEFSQc8eAfFfSvEn7LHjGbwvo3jq7urUxpeNPomoSw2kzuME jY+GYBQCfVcdqTSvitrH7LPxJ8daP4Xkha1uZ7rRJmvIhNJLbJK6bSRgAsAMkAHI4xXd/sif sieG/j58Nr7WNYvtctbq21OSzVbKaJIyixROCQ0bHOZD3xjHFfxHmmNwGKyrCYPL8P7KvHnv PnfR+9qtZcz1974do6H9p5fhMbhMzxeNzHEe1oS5LQ5I9V7uj0jyLT3W+bd6n//Z

--------------1jKwxgYezhBCiNfc2WH8HUwI--

bbtams commented 5 months ago

Hi Mickey, OFIQ implements what is specified in ISO/IEC 29794-5 (which is currently at DIS level). UQS is based on MagFace. MagFace does not necessarily check conformance but the mere utility of the image for being recognizable. There are no threshold specified in ISO/IEC 29794-5 yet. Best, Benjamin

P.S.: BTW it seems that you've confused the scores. For r-04 the UQS is 25 (not 78 which is background uniformity).