Skip to main content

A single library to (down)load all existing sign language handshape datasets. This library allows you to automatically download and load various sign language handshape datasets. Currently supporting 13 different datasets.

Project description

A single library to (down)load all existing sign language handshape datasets.

handshape handshape

There are various handshape datasets for Sign Language. However:

  • Each dataset has its own format and many are hard to find.
  • Each dataset has its own mapping of handshapes to classes. While Signs depend on the specific Sign Language for a country/region, handshapes are universal. Hence, they could be shared between datasets/tasks.

This library aims to provide two main features:

  • A simplified API to download and load handshape datasets
  • A mapping between datasets so that datasets can be merged for training/testing models.

We hope it will help Sign Language Recognition develop further, both for research and application development.

If you wish to add a dataset you can make a push request, file an issue, or write to handshape.datasets@at@gmail.

This library is a work in progress. Contributions are welcome.

Working with images

  • Identifiying Hand Classes
Letter Class ID
a 0
b 1
c 2
d 3
e 4
f 5
g 6
h 7
i 8
j 9

How to use?

Import handshape_datasets

handshape_datasets.load("dataset_id")

Download, extract and preprocess the dataset. The function will return "x" that contain an array with the images and metadata, this one contain an array with classes and if it have, an array with subjects or differents other values. For example, in lsa16 "x" will return a shape of (800,32,32,3). Also you could give a version value if its available to the selected dataset and you could give a boolean value to delete temporary files if its possible

Example:

handshape_datasets.load("lsa16",version="color",delete=True) --> download, extract and preprocess the lsa16 dataset in
version "color" and delete the temporary files if its have a .npz file.

handshape_datasets.clear("dataset_id") --> Delete all the local files for the dataset, if its exist.

handshape_datasets.list_datasets() --> Returns a table with the information for the availables datasets

handshape_datasets.delete_temporary_files("dataset_id") --> Delete the local files if its exist a .npz file

How to use the dataset in keras?

First, you must to load the dataset

dataset = handshape_datasets.load(dataset_id, version=ver, delete=supr)

You could have the input_shape and the number of classes

input_shape = self.dataset[0][0].shape
classes = self.dataset[1]['y'].max() + 1

Then you must to build a model

base_model = keras.applications.mobilenet.MobileNet(input_shape=(input_shape[0],self.input_shape[1],3), 
                                                            weights='imagenet', include_top=False)
output = keras.layers.GlobalAveragePooling2D()(base_model.output)
output = keras.layers.Dense(32, activation='relu')(output)
output = keras.layers.Dense(self.classes, activation='softmax')(output)
model = Model(inputs=base_model.input, outputs=output)
model.compile(optimizer='Adam', loss='sparse_categorical_crossentropy', metrics=['accuracy'])

Its optional to split the dataset for the validation data when fit the model

X_train, X_test, Y_train, Y_test = sklearn.model_selection.train_test_split(self.dataset[0], self.dataset[1]['y'],
                                                                                test_size=test_size,
                                                                                stratify=self.dataset[1]['y'])

At last you must to fit the model

history = model.fit(X_train, Y_train, batch_size=self.batch_size, epochs=self.epochs,
                             validation_data=(X_test, Y_test))

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

handshape-datasets-0.1.2.tar.gz (27.2 kB view details)

Uploaded Source

Built Distribution

handshape_datasets-0.1.2-py2.py3-none-any.whl (47.7 kB view details)

Uploaded Python 2 Python 3

File details

Details for the file handshape-datasets-0.1.2.tar.gz.

File metadata

  • Download URL: handshape-datasets-0.1.2.tar.gz
  • Upload date:
  • Size: 27.2 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/3.1.1 pkginfo/1.5.0.1 requests/2.23.0 setuptools/46.2.0 requests-toolbelt/0.9.1 tqdm/4.46.0 CPython/3.7.5

File hashes

Hashes for handshape-datasets-0.1.2.tar.gz
Algorithm Hash digest
SHA256 4b84d778d81f7a98aec8f47bd3698bc034e464c2b84187d462b58570450c2c72
MD5 88fb054d978baf8c4bb731455beb7dfa
BLAKE2b-256 eb57648e06391ec1492e525337268cd822a8025440490ddd9f896566c06bbfc5

See more details on using hashes here.

File details

Details for the file handshape_datasets-0.1.2-py2.py3-none-any.whl.

File metadata

  • Download URL: handshape_datasets-0.1.2-py2.py3-none-any.whl
  • Upload date:
  • Size: 47.7 kB
  • Tags: Python 2, Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/3.1.1 pkginfo/1.5.0.1 requests/2.23.0 setuptools/46.2.0 requests-toolbelt/0.9.1 tqdm/4.46.0 CPython/3.7.5

File hashes

Hashes for handshape_datasets-0.1.2-py2.py3-none-any.whl
Algorithm Hash digest
SHA256 653fd9c1e53f04ce879666a3412c814b17208962e8c19d8d4becb3081b561326
MD5 aaa458577f64cde5f048820e51f4342d
BLAKE2b-256 1743701b3ff33a19124be4fc1d71bddc81186b5189f55efdf99335f0d3596500

See more details on using hashes here.

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page