So I have been working on an object detection problem and I loaded my databunch object from the image list in fastai but I am not able to fully understand what kind of patches or images are genuinely present in my input data so I need a visualization of the data like how in what manner is it present how many positive classes, how much the data have hard negatives and their distributions. This is how I created my data bunch in fastai:
#here train_images,valid_images is a list of <object_detection_fastai.helper.wsi_loader.SlideContainer object>
batch_size = 64
do_flip = True
flip_vert = True
max_rotate = 90
max_zoom = 1.1
max_lighting = 0.2
max_warp = 0.2
p_affine = 0.75
p_lighting = 0.75
tfms = get_transforms(do_flip=do_flip,
flip_vert=flip_vert,
max_rotate=max_rotate,
max_zoom=max_zoom,
max_lighting=max_lighting,
max_warp=max_warp,
p_affine=p_affine,
p_lighting=p_lighting)
train, valid ,test = ObjectItemListSlide(train_images), ObjectItemListSlide(valid_images), ObjectItemListSlide(test_images)
#print("type",(test_set[12]))
item_list = ItemLists(".", train, test)
lls = item_list.label_from_func(lambda x: x.y, label_cls=SlideObjectCategoryList)
lls = lls.transform(tfms, tfm_y=True, size=patch_size)
data = lls.databunch(bs=batch_size, collate_fn=bb_pad_collate,num_workers=0).normalize()
Now I need to find out what patches or images are present. They are of which distribution like how many positives and how many negatives and background class, as after the transform function they are changed and randomized into the data bunch. Hence, the data bunch becomes a black box of my input data, I need to have insights on what is the actual distribution of data inside databunch.