Need help on making prediction with the model from this work.
yueyinlizun opened this issue · comments
Hi, I am planning to make some prediction about the CSI data collected by myself. Now I have restored a model graph from those three files, model.ckpt.data-00000-of-00001，model.ckpt.index，model.ckpt.meta. But I don't know how to make prediction. Can you give me some direction? Here is my code:
Sorry, I don't have enough resources to attack this problem.
Hope this link will help you.
https://stackoverflow.com/questions/33759623/tensorflow-how-to-save-restore-a-model
Thanks a lot for your help!
Hello, I have another problem and need your help. When I directly get the data from a .csv file, such as what about "walk", I get a numpy array with the shape of (*, 90000). I don't know what does the number 90000 mean, which is important for me to practically prepare some wifi activity data to make prediction. Waiting for your reply, Thank you!
Ummm I think my data shape is (~20000, 181).
20000 is the time (1ms sampling and 20 sec), 181 means (3 Antenna x 30 sub carrier x 2(amplitude & phase) + 1time).
Thanks for your clear explaination. I find the final input data required by the RNN model has a shape of (n, 500, 90), as shown in the picture below.
Does the number 90 just mean (3 antenna x 30 subcarrier x 1(amplitude or phase) ) ?
Yes!! 90 means (3 anntena x 30 subcarrier x 1 amplitude)!!
Oh! 90000 means 90 x 1000. 1000 is the window size for RNN.
I used 1 sec window size for RNN.
Thanks for your kindly reply. Thank you very much!
It's so kind of you. Can you tell me if you did some denoising preocess on those original data from .dat files before transforming them into .csv files? if so , what denoising function have you used?
Hello @yueyinlizun, did you ever figure out the answer to your question?
@yueyinlizun ,hi, I also want to make some prediction about the CSI data in the dataset. Have you finished the calling function? Thank you very much!