Chokkhu

Chokkhu is a deep learning image dataset EDA and preprocessing toolkit designed to prepare train-ready image data for TensorFlow / Keras–based image classification. It helps users analyze datasets, preprocess images, handle class imbalance, and train deep learning models using a clean and reproducible pipeline. The package follows industry-grade Python packaging standards, supports CI/CD pipelines, and works seamlessly in Google Colab and Jupyter Notebook environments.

INSTALLATION (IMPORTANT – FIRST STEP)

pip install Chokkhu <<<

TensorFlow is installed automatically as a runtime dependency.

Chokkhu’s main responsibility is data preparation. It performs image exploratory data analysis (EDA), class-wise distribution visualization, image size, aspect ratio, RGB intensity and blur analysis, standard preprocessing (resize to 224×224 and normalization), stratified train/validation/test splitting, and automatic class balancing using data augmentation. After this step, the dataset is fully ready to be used for training any deep learning model.

Complete usage example showing the full workflow in one place:

from Chokkhu import ImageEDA, ImagePreProcessor


# Dataset EDA
eda = ImageEDA(dataset_path="your_dataset_path")

# Dataset preprocessing
processor = ImagePreProcessor(datapath="your_dataset_path")
(train_X, train_y), (val_X, val_y), (test_X, test_y) = processor.get_data()




After excecuting this, You can train your model like this.







import tensorflow as tf
# Example 1: Custom CNN (from scratch)
custom_model = tf.keras.Sequential([
    tf.keras.layers.Conv2D(32, 3, activation="relu", input_shape=(224,224,3)),
    tf.keras.layers.MaxPooling2D(),
    tf.keras.layers.Conv2D(64, 3, activation="relu"),
    tf.keras.layers.MaxPooling2D(),
    tf.keras.layers.Flatten(),
    tf.keras.layers.Dense(128, activation="relu"),
    tf.keras.layers.Dense(num_classes, activation="softmax")
])

custom_model.compile(
    optimizer="adam",
    loss="sparse_categorical_crossentropy",
    metrics=["accuracy"]
)

custom_model.fit(
    train_X,
    train_y,
    validation_data=(val_X, val_y),
    epochs=10
)




# Example 2: Transfer Learning with ConvNeXt-Tiny (frozen backbone)
import tensorflow as tf
base_model = tf.keras.applications.ConvNeXtTiny(
    weights="imagenet",
    include_top=False,
    input_shape=(224,224,3)
)

base_model.trainable = False

transfer_model = tf.keras.Sequential([
    base_model,
    tf.keras.layers.GlobalAveragePooling2D(),
    tf.keras.layers.Dense(256, activation="relu"),
    tf.keras.layers.Dense(num_classes, activation="softmax")
])

transfer_model.compile(
    optimizer="adam",
    loss="sparse_categorical_crossentropy",
    metrics=["accuracy"]
)

transfer_model.fit(
    train_X,
    train_y,
    validation_data=(val_X, val_y),
    epochs=5
)




# Example 3: Fine-tuning ConvNeXt-Tiny (unfrozen backbone)
import tensorflow as tf
base_model.trainable = True

transfer_model.compile(
    optimizer=tf.keras.optimizers.Adam(1e-5),
    loss="sparse_categorical_crossentropy",
    metrics=["accuracy"]
)

transfer_model.fit(
    train_X,
    train_y,
    validation_data=(val_X, val_y),
    epochs=5
)

Metadata

Release files for chokkhu 0.3.4

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for chokkhu 0.3.4
File	Size	Uploaded
chokkhu-0.3.4.tar.gz	17.8 kB	Aug 1, 2026	Details

Built distribution (wheel)

Table of built distributions (wheels) for chokkhu 0.3.4
File	Interpreter	ABI	Platform	Reset
chokkhu-0.3.4-py3-none-any.whl 20.9 kB Aug 1, 2026	Python 3	none	any	Details

Total release size: 38.7 kB

Release files / chokkhu-0.3.4.tar.gz

Download URL	chokkhu-0.3.4.tar.gz
Size	17.8 kB
Tags	Source
SHA-256 checksum How to use checksums	`bcad9a29ddef5d19d0c11a8895fbeb74a2bdaaf8b6598fc997a49d9836e8fdc9`
BLAKE2b-256 checksum How to use checksums	`51051a0a08cca298401f60b3b6ec0d310f20c673417ac690a7101cec42fc8ccc`
Upload date	Aug 1, 2026
Uploaded using Trusted Publishing? What is trusted publishing?	No
Uploaded via	`twine/7.0.0 CPython/3.13.14`

Release files / chokkhu-0.3.4-py3-none-any.whl

Download URL	chokkhu-0.3.4-py3-none-any.whl
Size	20.9 kB
Tags	Python 3
SHA-256 checksum How to use checksums	`d410e3083abf38a114589c76336e3effd3f84ed6c7c50e710a1b709c65618488`
BLAKE2b-256 checksum How to use checksums	`6290e484a49e20f50d43e73bf92669960f2cf73c5fc1a7ec1bceca3b853bbfd5`
Upload date	Aug 1, 2026
Uploaded using Trusted Publishing? What is trusted publishing?	No
Uploaded via	`twine/7.0.0 CPython/3.13.14`