chitra 0.0.4

Library for creating data input pipeline in pure Tensorflow 2.x

Chitra

Library for creating data input pipeline in pure Tensorflow 2.x

.

Install

pip install chitra

How to use

Loading data for image classification

import tensorflow as tf
import chitra
from chitra.dataloader import Clf, show_batch

path = '/Users/aniketmaurya/Pictures/cats'

clf_dl = Clf()
data = clf_dl.from_folder(path)

print('class names:', clf_dl.CLASS_NAMES)

show_batch(data, 6, (6, 6))

class names: ('Whitecat', 'Blackcat')

model = tf.keras.applications.ResNet50(include_top=False,
                                              weights='imagenet',
                                              input_shape=(160, 160, 3),
                                              classes=2)

model.fit(data)

---------------------------------------------------------------------------

RuntimeError                              Traceback (most recent call last)

<ipython-input-7-bcfd88bde046> in <module>
----> 1 model.fit(data)


~/miniconda3/envs/tf/lib/python3.7/site-packages/tensorflow_core/python/keras/engine/training.py in fit(self, x, y, batch_size, epochs, verbose, callbacks, validation_split, validation_data, shuffle, class_weight, sample_weight, initial_epoch, steps_per_epoch, validation_steps, validation_freq, max_queue_size, workers, use_multiprocessing, **kwargs)
    703     if kwargs:
    704       raise TypeError('Unrecognized keyword arguments: ' + str(kwargs))
--> 705     self._assert_compile_was_called()
    706     self._check_call_args('fit')
    707 


~/miniconda3/envs/tf/lib/python3.7/site-packages/tensorflow_core/python/keras/engine/training.py in _assert_compile_was_called(self)
   2872     # (i.e. whether the model is built and its inputs/outputs are set).
   2873     if not self.optimizer:
-> 2874       raise RuntimeError('You must compile your model before '
   2875                          'training/testing. '
   2876                          'Use `model.compile(optimizer, loss)`.')


RuntimeError: You must compile your model before training/testing. Use `model.compile(optimizer, loss)`.

# for e in data.batch(4): print(e)

img = chitra.image.read_image('/Users/aniketmaurya/Pictures/cats/whitecat/wcat1.jpg')
img.shape

TensorShape([683, 1024, 3])

chitra.image.resize_image(img, (160, 160)).shape

TensorShape([160, 160, 3])