CaptainAshis

! unzip labels.csv.zip 
! unzip sample_submission.csv.zip
! unzip test.zip
! unzip train.zip

CaptainAshis

labels_df=pd.read_csv(f'{PATH}labels.csv')
labels_df.head()

CaptainAshis

label_csv=f'{PATH}labels.csv'
n=len(list(open(label_csv)))-1
val_idxs=get_cv_idxs(n)

CaptainAshis

print("Total size",n)
print("Total validation size",len(val_idxs))

####################
#Total size 10222
#Total validation size 2044

CaptainAshis

sz=224 # size of image
arch=resnext101_64 # architecture used 
bs=64 # batch size used

CaptainAshis

data=  ImageClassifierData.from_csv(PATH,'train',f'{PATH}labels.csv',test_name='test',
                                     val_idxs=val_idxs,suffix='.jpg',tfms=tfms_from_model(arch, sz),bs=64)

CaptainAshis

fn=PATH+data.trn_ds.fnames[1]
fn

## '/content/clouderizer/dogbreed/data/competitions/dog-breed-identification/train/001cdf01b096e06d78e9e5112d419397.jpg'

img=PIL.Image.open(fn)
img

CaptainAshis

img.size

##(500, 375)

# Mapping name of the file to size of the file
size_d={k:PIL.Image.open(PATH+k).size for k in data.trn_ds.fnames}

row_sz,col_sz=list(zip(*size_d.values())) 

row_sz=np.array(row_sz)

CaptainAshis

len(data.trn_ds),len(data.test_ds)

### (8178, 10357)

len(data.classes),data.classes[:5]

##################################
# (120,
#  ['affenpinscher',
#   'afghan_hound',

CaptainAshis

def get_data(sz,bs):
    tfms=tfms_from_model(arch, sz,aug_tfms=transforms_side_on,max_zoom=1.1)
    data=  ImageClassifierData.from_csv(PATH,'train',f'{PATH}labels.csv',test_name='test',num_workers=8,
                                     val_idxs=val_idxs,suffix='.jpg',tfms=tfms,bs=bs)
    return data if sz>300 else data.resize(340,'tmp')
    
 data=get_data(sz,bs)