Python [/code] code implementation judge of adjacent images or continuous picture
A total of 500 images, there is a TXT documents, document is inside the picture information but not all image contains only part of the information,
For no image TXT document information you need to determine the current image quality,
Known every two images must have a is good
For example: 5,6,8,9,5689 is the number of the picture, zhang no 7 TXT documents information then this is a good
3 and 7, 3 and 7 is the picture name, between 3 and 7, no. 4 and 6 images without TXT message, so there must be some good just didn't save in the 4 and 6 information, so you need to determine which is a good 4 and 6?

CodePudding user response:

Sensory information is less, the basis of a judge pictures only of adjacent images, good or bad? Every two images must have a is good, that means at least one is good, can't be bad for two, but the first case, you did not show the stand or fall of 6 and 8 information is directly decided that 7 is good, if 6 and 8 are good, then 7 is bad also no problem still accord with what you say every two images must have a is good, and so to you the proposed conditions will appear a lot of kinds of situation, can not only determine the stand or fall of image,

CodePudding user response:

reference 1st floor Dihambo response:

feel less information, judge images based on only the stand or fall of adjacent images? Every two images must have a is good, that means at least one is good, can't be bad for two, but the first case, you did not show the stand or fall of 6 and 8 information is directly decided that 7 is good, if 6 and 8 are good, then 7 is bad also no problem still accord with what you say every two images must have a is good, and so to you the proposed conditions will appear a lot of kinds of situation, can not only determine the stand or fall of image,

Thank you for your reply, is my expression has a problem, so, I have a batch of image target detection using neural network to run, such as 1. 0.98 JPG, (2) 0.78 JPG, 6. 0.80 JPG (0.98, 0.78, 0.80 is a target in the image recognition rate), I want to know 3. JPG, 4. JPG, 5. JPG inside have a goal, so sure that misses or not, I calculated the distance of adjacent target in the image pixel is 1200 px - 2000 px, want to know how to use the code to implement such a situation?

CodePudding user response:

references a learn 2 floor response:

Quote: refer to 1st floor Dihambo response:

Sensory information is less, the basis of a judge pictures only of adjacent images, good or bad? Every two images must have a is good, that means at least one is good, can't be bad for two, but the first case, you did not show the stand or fall of 6 and 8 information is directly decided that 7 is good, if 6 and 8 are good, then 7 is bad also no problem still accord with what you say every two images must have a is good, and so to you the proposed conditions will appear a lot of kinds of situation, can not only determine the stand or fall of image,

Thank you for your reply, is my expression has a problem, so, I have a batch of image target detection using neural network to run, such as 1. 0.98 JPG, (2) 0.78 JPG, 6. 0.80 JPG (0.98, 0.78, 0.80 is a target in the image recognition rate), I want to know 3. JPG, 4. JPG, 5. JPG inside have a goal, so sure that misses or not, I calculated the distance of adjacent target in the image pixel is 1200 px - 2000 px, want to know how to use the code to implement such a situation?

Then you put the correct recognition rate and the target recognition, to see what reason be

CodePudding user response:

reference weixin_45903952 reply: 3/f

Quote: refer to the second floor a learn response:

Quote: refer to 1st floor Dihambo response:

Sensory information is less, the basis of a judge pictures only of adjacent images, good or bad? Every two images must have a is good, that means at least one is good, can't be bad for two, but the first case, you did not show the stand or fall of 6 and 8 information is directly decided that 7 is good, if 6 and 8 are good, then 7 is bad also no problem still accord with what you say every two images must have a is good, and so to you the proposed conditions will appear a lot of kinds of situation, can not only determine the stand or fall of image,

Thank you for your reply, is my expression has a problem, so, I have a batch of image target detection using neural network to run, such as 1. 0.98 JPG, (2) 0.78 JPG, 6. 0.80 JPG (0.98, 0.78, 0.80 is a target in the image recognition rate), I want to know 3. JPG, 4. JPG, 5. JPG inside have a goal, so sure that misses or not, I calculated the distance of adjacent target in the image pixel is 1200 px - 2000 px, want to know how to use the code to implement such a situation?

Then you put the correct recognition rate and the target recognition, to see what reason be

Well, I think of the problem of the precision of the neural network can't reach 100%, so there is part of the testers, you excuse me, how do I use program to realize the residual image screening out? Like 1. JPG, 2. JPG, 6. JPG three, four, five, in the middle of the picture is because the network precision insufficient lead to leak, I need to take these to find and do some markers, such as adding a picture on the left or the right red color bar, just like the image below