X,y=next(iter(md.aug_dl.dataset)) return 0's in y

(Teja) #1

Hello All! I’m working on a problem similar to lesson 8. I’m trying to predict the bounding boxes of images. While using data augmentation, I’m using this
augs = [RandomFlip(tfm_y=TfmType.COORD),
RandomRotate(30, tfm_y=TfmType.COORD),
RandomLighting(0.1,0.1, tfm_y=TfmType.COORD)]
and

tfms = tfms_from_model(f_model, sz, crop_type=CropType.NO, aug_tfms=augs)
md = ImageClassifierData.from_csv(PATH, JPEGS, BB_CSV, tfms=tfms, continuous=True, bs=4)

as mentioned in pascal notebook of lesson 8,

On running this

x,y=next(iter(md.aug_dl.dataset))

I get all zeros in y. The output is as below

0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
[torch.cuda.FloatTensor of size 4x4 (GPU 0)]

I checked y values by running

md.aug_dl.dataset.y whcih gives

array([[139., 552., 139., 413.],
[111., 500., 160., 347.],
[154., 515., 170., 343.],
…,
[251., 381., 124., 308.],
[258., 387., 40., 351.],
[258., 374., 64., 434.]], dtype=float32)

But I don’t understand why it returns zeros in y when I run
x,y=next(iter(md.aug_dl.dataset))

I’m stuck here. Please, help me out!

Any help will be hugely appreciated.

0 Likes

(Shreyas S) #2

I have the same exact problem! Did you find a solution?

0 Likes