Open Phoebe-star opened 7 years ago
Replace "your_dataset_path" to your downloaded dataset path. For example "/tmp/cuhk03".
thanks , I solved but how to do input two images and get the result whether if they are the same or not?
I added test function. Refer to the README.
thanks , I have another problem when training , how to know the two input image is the same or not the same person
I make True/False pair from dataset repeatedly. Refer the 'read_data' function in 'cuhk03_dataset.py'.
Does not check the image name when running the run.py ? like 0000_00.jpg and 0000_01.jpg is the same 0000_01.jpg and 0001_00.jpg is not the same
can it train own data?how to do? thanks you^^
Maybe false prediction. Accuracy is not 100% :)
I used CUHK03 dataset. It is composed of mat files. If your dataset is different format, you should make your own dataset.py.
I have some questions?
in the cuhk03_dataset.py , line 79 and line80 ( labels.append([1., 0.]) labels.append([0., 1.]) ) Q: what is doing??
in the cuhk03_dataset.py , line90 ( return np.transpose(batch_images, (1, 0, 2, 3, 4)) )
Q:why is 1,0,2,3,4 instead of 0,1,2,3,4
3.in the run.py line 84 and line85 reshape = tf.reshape(concat, [shape[2], shape[3], shape[0], shape[1], 5, 5]) g = tf.transpose(reshape, [2, 3, 0, 1, 4, 5]) Q: why is 2 ,3, 0 , 1 and 2,3,0,1,4,5
4.do yo have any idea for the thing ?
the thing is when training , I use the reverse picture (the person head is button, the person leg is top)
it will happen?
and if I train the picture that the WIDTH is longer than HEIGHT
can it train?
thank you very much
@thomaspark-pkj hello, I have run like this:python run.py --data_dir=/home/dai/code/person_reidentification/5/cuhk03_release but I still got a problem that :TypeError: split() takes at least 3 arguments (3 given). Could you please tell me how to solve it?
@thomaspark-pkj hello I solved the problem,and run the code. But I also want to show the feature like the visualization of feature in the paper,I run the code and find the feature in the conv layers or other layers is four dimensions (1,37,12,25) that is not the three dimensions that is RGB,so I cannot get the result with the color like the visualization of feature as shown in your paper. Could you please tell me what is wrong? Thankyou verymuch
@sde123 hello, I have ran: python run.py and got the same problem: TypeError: split() takes at least 3 arguments (3 given). I wonder how did you solve it? ths!
My code:
filedata = database.read() database.close() newdata=filedata.replace(userStuff[0]+" "+userStuff[1]+" "+userStuff[2]+userStuff[0]+" "+userStuff[1]+" "+str(player1score)) database=open("usernames","w") database.write(newdata)
it came up with this error: TypeError: replace() takes at least 3 arguments (2 given)
I am doing it in trinket and wondered how to fix it
I run the command "python run.py --data_dir=your_dataset_path" or
"python run.py --mode=val --data_dir=your_dataset_path"
and have this error
Traceback (most recent call last): File "run.py", line 183, in
tf.app.run()
File "/home/EN/person_search/local/lib/python2.7/site-packages/tensorflow/python/platform/app.py", line 43, in run
sys.exit(main(sys.argv[:1] + flags_passthrough))
File "run.py", line 121, in main
images1, images2 = preprocess(images, is_train)
File "run.py", line 48, in preprocess
return tf.cond(is_train, train, val)
File "/home/EN/person_search/local/lib/python2.7/site-packages/tensorflow/python/ops/control_flow_ops.py", line 1776, in cond
orig_res, res_t = context_t.BuildCondBranch(fn1)
File "/home/EN/person_search/local/lib/python2.7/site-packages/tensorflow/python/ops/control_flow_ops.py", line 1677, in BuildCondBranch
r = fn()
File "run.py", line 19, in train
split = tf.split(images, [1, 1])
TypeError: split() takes at least 3 arguments (2 given)