How to visualize other datasets test set pictures?

wzzhai commented 2 years ago

Thanks for the author's impressive work! Question 1: How to visualize test set pictures for SHHA, SHHB, UCF-QNRF, FDST?

Question 2: The link (DATA.md) is broken, could you please update the link, especially SHHA and SHHB, thanks.

taohan10200 commented 2 years ago

A1: You can generate the prediction results for other students, and then modify the vis4val .py to generate the visual results. A2: We have created another download link for the Shanghai Tech Dataset.

wzzhai commented 2 years ago

Thank you for your reply in your busy schedule. May I ask you three questions?

Question 1：

The txt files of the test set SHHB_VGG16_FPN_test.txt and SHHB_HR_Net_test.txt are used in the test set of SHHB provided by the author. The code is as follows:

dataset = 'SHHB'
dataRoot = '/Documents/Datasets/IIM/ProcessedData/' + dataset
gt_file = dataRoot + '/test_gt_loc.txt'
img_path = ori_data = dataRoot + '/images'

flagError = False
id_std = [i for i in range(401,716,1)]
# id_std[59] = 3098

exp_name = './saved_exp_results/SHHB_vis_results'
pred_file = '/Documents/Datasets/IIM/PredictionsOnTestSet/SHHB_VGG16_FPN_test.txt'

Prompt me error :

401
Traceback (most recent call last):
  File "vis4val.py", line 129, in <module>
    main()
  File "vis4val.py", line 125, in main
    cv2.imwrite(exp_name+'/'+str(i_sample)+ '_pre_' + str(pre)[0:6] + '_rec_' + str(rec)[0:6] + '.jpg', img)
cv2.error: OpenCV(3.4.11) /tmp/pip-req-build-1cb5e5yp/opencv/modules/imgcodecs/src/loadsave.cpp:759: error: (-215:Assertion failed) !_img.empty() in function 'imwrite'

The same error occurs in the same dataset SHHA as above. Could you please ask how to adjust the code?

Question2：

JHU's test set image order is random, id_std=? . Could you please ask how to adjust the code? The code is as follows:

dataset = 'JHU'
dataRoot = '/Documents/Datasets/IIM/ProcessedData/' + dataset
gt_file = dataRoot + '/test_gt_loc.txt'
img_path = ori_data = dataRoot + '/images'

exp_name = '/Experiments/IIM/IIM_HR/saved_exp_results/VGG_JHU_vis_results'
pred_file = '/Documents/Datasets/IIM/PredictionsOnTestSet/JHU_VGG16_FPN_test.txt'

flagError = False
id_std = [i for i in range(2409,4380,1)]

Prompt me error :

2409
Traceback (most recent call last):
  File "vis4val.py", line 129, in <module>
    main()
  File "vis4val.py", line 125, in main
    cv2.imwrite(exp_name+'/'+str(i_sample)+ '_pre_' + str(pre)[0:6] + '_rec_' + str(rec)[0:6] + '.jpg', img)
cv2.error: OpenCV(3.4.11) /tmp/pip-req-build-1cb5e5yp/opencv/modules/imgcodecs/src/loadsave.cpp:759: error: (-215:Assertion failed) !_img.empty() in function 'imwrite'
KeyError: 2410

And what is id_std=? for FDST?

dataset = 'FDST'
dataRoot = '/bulk/wenzhe/Documents/Datasets/IIM/ProcessedData/' + dataset
gt_file = dataRoot + '/test_gt_loc.txt'
img_path = ori_data = dataRoot + '/images'

exp_name = '/Experiments/IIM/IIM_HR/saved_exp_results/VGG_FDST_vis_results'
pred_file = '/Documents/Datasets/IIM/PredictionsOnTestSet/FDST_HR_Net_test.txt'

flagError = False
id_std = [i for i in range(10_001,99_150,1)]

Prompt me error, import data 10001 instead of 10001 , Could you please tell me how to modify the code?

10001
Traceback (most recent call last):
  File "vis4val.py", line 148, in <module>
    main()
  File "vis4val.py", line 79, in main
    if gt_data[i_sample]['num'] ==0 and pred_data[i_sample]['num'] !=0:

Question3：

When the training set is set to NWPU, an error will be reported. Could you please tell me what caused it and how to modify the code? Thanks.

====================
  0%|                                                                                           | 0/500 [00:00<?, ?it/s]Traceback (most recent call last):
  File "train.py", line 31, in <module>
    cc_trainer.forward()
  File "/bulk/wenzhe/Experiments/IIM/IIM_HR/trainer.py", line 72, in forward
    self.validate()
  File "/bulk/wenzhe/Experiments/IIM/IIM_HR/trainer.py", line 183, in validate
    assert img.size()[2:] == dot_map.size()[2:]
AssertionError

Looking forward to your reply. Thank you very much for your kind help.

taohan10200 commented 2 years ago

Q1: you can try to fix it by installing the opencv-python package, here is my opencv version:

opencv-contrib-python==4.2.0.34
opencv-python==4.5.1.48

Q2: One way to generate the id list for JHU is to read the txt list of JHU test set, like bellow 2409 0844 0400 0700 1872 2985 2873 1563 3987 3603 3551 0651 3255 3148 2946 ....

We here only provide the code for the visualization of several typical datasets. This code would be failed when the data format is not the same as the code targeted. For FDST's visualization, you can figure it out after you know how the code works.

Q3: Have you generated the dot map correctly? it is recommended to debug step by step and find why your image size is not the same as the dot_map.

wzzhai commented 2 years ago

Thanks your patient explanation extremely ！ It was replaced according to your instructions opencv-contrib-python and opencv-python, but the error cv2.error: OpenCV(4.5.4) still exists. After verification, it is found that the file name may be the reason. For the SHHB data set, the same error is displayed when 'id_std = [I for I in range(401,716,1)]'. However, renaming the picture0401.jpg to 401.jpg, and replacing all 0401 in the .txt with 401 only works for one 401.jpg picture. So is there any way to run this code besides changing the way of renaming?

I replace 401 with 0401, and replaced the code with the following:

    id_std = [i for i in range(0401,0716,1)]

or

    id_std = [i for i in range(''0401'',''0716'',1)]

The error is:

  File "vis4val.py", line 54
    id_std = [i for i in range(0401,0716,1)]
                                  ^
SyntaxError: invalid token

and

  File "vis4val.py", line 54
    id_std = [i for i in range(''0401'',''0716'',1)]
                                    ^
SyntaxError: invalid token

Could you please tell me what caused it and how to modify id_std = ? or modify other code ? Thanks.

Thank you for writing back to me from your busy schedule.

taohan10200 commented 2 years ago

id_stdis an id list of the pictures that also have been saved as keys in test_gt_loc.txt, id list is set in advance to index the prediction result and the ground truth result for each image from the gt_loc.txt and pred_loc.txt.

[i for i in range(''0401'',''0716'',1)] or id_std = [i for i in range(0401,0716,1)] are the wrong usages cause you can't range between two strings. They must be integer.

wzzhai commented 2 years ago

Thank you very much for your reply！

cyn2399 commented 6 months ago

Thanks your patient explanation extremely ！ It was replaced according to your instructions opencv-contrib-python and opencv-python, but the error cv2.error: OpenCV(4.5.4) still exists. After verification, it is found that the file name may be the reason. For the SHHB data set, the same error is displayed when 'id_std = [I for I in range(401,716,1)]'. However, renaming the picture0401.jpg to 401.jpg, and replacing all 0401 in the .txt with 401 only works for one 401.jpg picture. So is there any way to run this code besides changing the way of renaming?

I replace 401 with 0401, and replaced the code with the following:
    id_std = [i for i in range(0401,0716,1)]
or
    id_std = [i for i in range(''0401'',''0716'',1)]
The error is:
  File "vis4val.py", line 54
    id_std = [i for i in range(0401,0716,1)]
                                  ^
SyntaxError: invalid token
and
  File "vis4val.py", line 54
    id_std = [i for i in range(''0401'',''0716'',1)]
                                    ^
SyntaxError: invalid token
Could you please tell me what caused it and how to modify id_std = ? or modify other code ? Thanks.

Thank you for writing back to me from your busy schedule.

About Q1, you can try fix this code. img = cv2.imread(img_path + '/' + str(i_sample).zfill(4) + '.jpg')

taohan10200 / IIM