UIED - UI element detection part of UI2CODE, detecting UI elements from UI screenshots or drawnings

This project is still ongoing and this repo may be updated irregularly, I also implement a web app for this project in http://uied.online

Related Publications:

1. UIED: a hybrid tool for GUI element detection

2. Object Detection for Graphical User Interface: Old Fashioned or Deep Learning or a Combination?

What is it?

UI Element Detection (UIED) is an old-fashioned computer vision (CV) based element detection approach for graphic user interface.

The input of UIED could be various UI image, such as mobile app or web page screenshot, UI design drawn by Photoshop or Sketch, and even some hand-drawn UI design. Then the approach detects and classifies text and graphic UI elements, and exports the detection result as JSON file for future application.

UIED comprises two parts to detect UI text and graphic elements, such as button, image and input bar.

• For text, it leverages a state-of-the-art scene text detector EAST to perfrom detection.

• For graphical elements, it uses old-fashioned CV and image processing algorithms with a set of creative innovations to locate the elements and applies a CNN to achieve classification.

How to use?

Dependency

• Python 3.5
• Numpy 1.15.2
• Opencv 3.4.2
• Tensorflow 1.10.0
• Keras 2.2.4
• Sklearn 0.22.2
• Pandas 0.23.4

Installation

Install the mentioned dependencies, and download two pre-trained models from this link for EAST text detection and GUI element classification.

Change CNN_PATH and EAST_PATH in config/CONFIG.py to your locations.

Usage

To test your own image(s):

• For testing single image, change input_path_img in run_single.py to your own input image and the results will be outputted to output_root.
• For testing mutiple images, change input_img_root in run_batch.py to your own input directory and the results will be outputted to output_root.

Note: The best set of parameters vary for different types of GUI image (Mobile App, Web, PC). Three of critical ones are {'param-grad', 'param-block', 'param-minarea'} which can be easily adjusted in detect_compo\ip_region_proposal.py.

File structure

cnn/

• Used to train classifier for graphic UI elements
• Set path of the CNN classification model

config/

• Set data paths
• Set parameters for graphic elements detection

data/

• Input UI images and output detection results

detect_compo/

• Graphic UI elemnts localization
• Graphic UI elemnts classification by CNN

detect_text_east/

• UI text detection by EAST

result_processing/

• Result evaluation and visualizition

merge.py

• Merge the results from the graphical UI elements detection and text detection

run_batch.py

• Process a batch of images

run_single.py

• Process a signle image

Demo

GUI element detection result for web screenshot