A.I & Data Science Challenges | DPhi

Problem Discussion
#1
This is a companion discussion topic for the original entry at https://dphi.tech/challenges/144
#2

Can anyone help me? I dont understand what are we supposed to do in preprocessing part of image data? Its already in 100x100 size. Do I need to flatten it into 100x1 shape?

#3

normalize by dividing by 255.0

1 Like
#4

how to find the source of train images for cv2.resize??
I am getting the error (-215:Assertion failed) !ssize.empty() in function ‘cv::resize’
again and again.

#5

How many submissions can we make?
If multiple submissions are possible, Is it like we will have to upload notebook for every submission and then would have the choice to choose best score notebook?

#6

The size of each image is different any idea what can I do with it?

#7

@kshitizdoon there is no submission limit. Also, you can upload the notebook when you are done with the iterations i.e you don’t need to upload it on each and every submission.

1 Like
#8

@kashish_ranjan you must normalize the images and bring every image to same shape. You will find those tips in the starter notebook here: DPhi - Notebooks

Also, this was taught in one of the sessions, please feel free to refer the learning material that is put up the on the bootcamp resources.

#9

@vinyk.sd could you please check the below two resources, it might be handy.

#10

Hi @crimsondawn7

We have provided a basic steps with how to load and read the data.
There are some other preprocessing steps like one mentioned by @emma to normalize the pixel values. you can also further resize the array of the images accordingly as per the requirement or perform data augmentation, etc.

#11

Hi @vinyk.sd
this issue might also arise if you are not reading the data properly. If no image is read from the given location (i.e. the image array is empty), cv resize will not work.

#12

Screenshot (143)
Screenshot (143)1920×1080 226 KB

is this how my data should look like?

#13

@manish_kc_06 @chanukya pls let me know once.I am confused with this

#14

There is less time left kindly help me out with this.

#15

Is there any certain file names of the notebook for the bootcampers that I should follow ?

#16

hey, you are facing this issue because you haven’t read the train folder(image folder) properly , i also faced the same issue in beginning just use '/ ’ in the end of the path of train folder.
i hope this will help uh

#17

there is nothing wrong in this interpretation

#18

how do i split this data?

#19

For some reason which i dont understand…im not able to predict values using my model. As soon as i make a presiction, all the values come out to be 0. i dont understand why this is happening.

#20

My model is only predicting one class (1) for male. Is this a result of a lack of data, such that I need to increase training data via data augmentation?