title
SAS Deep Learning Python (DLPy)
shelf-menu
title
SAS Deep Learning Python (DLPy)
description
DLPy is a high-level Python library for the SAS Deep Learning features available in SAS® Viya®. DLPy provides a convenient way to apply deep learning functionalities to solve computer vision, NLP, forecasting and speech processing problems.
background-image

DLPy features include the following:

  • Read in and build deep learning models for image, text, audio and time series data.

    • High-level APIs for:
    • Deep neural networks for tabular data.
    • Image classification and regression.
    • Object detection.
    • RNN-based tasks – text classification, text generation and sequence labeling.
    • RNN-based time series processing and modeling.
  • Processing audio files and training deep learning algorithms to create a language model for speech recognition applications.
  • Predefined network architectures such as LeNet, VGG, ResNet, DenseNet, Darknet, Inception and YoloV2 and Tiny_Yolo. And many are provided with pre-trained weights!
  • Enhanced data visualization such as heat maps and feature maps to aid in the interpretation of deep learning computer vision models.
  • Import and export deep learning models in ONNX format.

DLPy can be used from Jupyter Notebook, JupyterLab, or from any Python console/scripting environment.

DLPy is open source and available on GitHub and PyPI. User contributions are accepted.

heading

Get Started

text-align
center

Find all DLPy pre-requisites, installation, and configuration on sassoftware.github.io/python-dlpy/.

DLPy can be installed using pip as follows:
pip install sas-dlpy
If you use the Anaconda distribution of Python, you can also use:
conda install -c sas-institute sas-dlpy

You can also download from the GitHub repository and install locally.

Deep Learning with Python (DLPy) Computer Vision Demo Series

Introduction to the DLPy computer vision video series

Introduction to the DLPy computer vision video series

Leverage the Functional API with DLPy to build complex models for feature extraction, including a ResNet model

Leverage the Functional API with DLPy to build complex models for feature extraction, including a ResNet model

Image segmentation is achieved with a U-Net model to track soccer players

Image segmentation is achieved with a U-Net model to track soccer players

DLPY supports object detection with Faster R-CNN using VGG16 as the backbone network

DLPY supports object detection with Faster R-CNN using VGG16 as the backbone network

DLPy supports MobileNet and ShuffleNet and compares the performance of those models with a ResNet50 model

DLPy supports MobileNet and ShuffleNet and compares the performance of those models with a ResNet50 model

Multi-task learning is performed for training in the retail domain for product recommendations

Multi-task learning is performed for training in the retail domain for product recommendations

Multi-task learning is performed for inferencing in the retail domain for product recommendations

Multi-task learning is performed for inferencing in the retail domain for product recommendations

DLPy and SAS® Viya® Demo Resources

DLPy and SAS Viya - Introduction

Introduction

DLPy and SAS Viya - Image Classification with Convolutional Neural Networks

Image Classification with Convolutional Neural Networks

DLPy and SAS Viya - Obj Detection With Tiny YOLOv2 Model

Obj Detection With Tiny YOLOv2 Model

DLPy and SAS Viya - Import and Export Deep Learning Models With ONNX

Import and Export Deep Learning Models With ONNX

DLPy and SAS Viya - Text Class. and Text Generation With Recurrent Neural Networks

Text Class. and Text Generation With Recurrent Neural Networks

DLPy and SAS Viya - Time Series Forecasting With Recurrent Neural Networks

Time Series Forecasting With Recurrent Neural Networks

DLPy Documentation

heading

Samples

text-align
center

Example Notebooks and use cases can be found at sassoftware/python-dlpy/examples.

Sample Notebook Screenshots

Heat maps generated using the score results from a Convolutional Neural Network (CNN) model
DLPy model built using the Functional API
Detected objects from table images enclosed in bounding boxes
One-step-ahead time-series forecast
heading

Resources

subheading
Conference papers and presentations
text-align
center
heading

DLPy Blogs & Communities

text-align
center

DLPY Blogs

SAS Support Communities

Developers Community

Data Mining and Machine Learning Community

DLPy Community Articles