Hi everyone,
I’m working on semantic segmentation. For training and testing data, I created .csv file which contains image names. When I’m training to transform data I get error
Exception: It’s not possible to apply those transforms to your dataset:
‘list’ object has no attribute ‘name’.
Here is my code:
> get_y_fn = lambda x: (str(x).replace("image_patches", "mask_patches"))
>
> test_data = ImageList.from_csv(path=path_img, csv_name="04_test.csv",header="infer")
>
> src = (SegmentationItemList.from_csv(path=path, csv_name='04_train.csv', folder='image_patches')
> .split_by_rand_pct(0.2)
> .label_from_func(get_y_fn, classes=codes)
> .add_test(test_data, tfm_y=False))
> tfms = get_transforms(do_flip=True, flip_vert=True, max_warp=0, max_zoom=1.2, max_lighting=0.3, max_rotate=359)
>
> data = (src.transform(tfms, size=size, tfm_y=True)
> .databunch(bs=bs)
> .normalize(imagenet_stats))