Hi guys, i am going through the image segmentation part of lesson 3 (2019).
One thing i didn’t understand is: once i trained my model, can i run a normal picture as a test and see which type of labels i see inside?
Imagine i have a dataset of fashion clothes segmented that i usw to train the model to detect fashion details.
After training, can i send a normal picture in the model or this one has to be segmented also?
I am lookibg to create a multilabel fashion image detector and would like to know if segmentation could be a solution.
What do you mean by a normal image? You can pass any image to your model as input.
Like, the image i use to test the model has to be segmented also? Or i just can send in any picture i find online?
Can i somehow get a multilabel classification on a picture out of a segmentation model?
If you are training it with image + segmented image (mask), then yes for prediction you will need the same. If you are training an image to give the segmentation as the output, normal image is fine.
One option is have 2 models, one train it to generate segmentation from image and then another train it to detect fashion details from segmentation. That way, you can input a normal image, get the segmentation from your first model and finally get details from second. Another option is have a single model output both things (a bit harder) and another is to not use segmented information.
thanks a lot for the reply! got it
The below is for image segmentation :
- How to get the color vs class mapping.
- what is the Difference between pred_class and pred_class.show(), the see the mask colour changes b/w the two
ip='/content/gdrive/My Drive/Colab Notebooks/Camvid/2.jpg'
ipg = open_image(ip)
pred_class,pred_idx,outputs = learn.predict(ipg)
for i in pred_idx:
for i in l:
Building , 4
Car , 5
ParkingBlock , 15
Road , 17
Sidewalk , 19
Sky , 21
SUVPickupTruck , 22
Tree , 26
Void , 30