VID2XML: Automatic Extraction of a Complete XML Data from Mobile Programming Screencasts

0
0 ratings - Please login to submit your rating.

Abstract 

TO BE ADDED AFTER PUBLICATION. 

Instructions: 

#The replication contains different folder which we explain as follows. 

#phase1 
This folder contains the visual results of words, lines, and initial XML bounding boxes detection for the 10 different videos named as (vidID_debugs). 
Each image is named as (frameNumber_tagName), in which tagName could be words, lines, combined (words+lines), and bb (which shows a predicted boudning box for the XML region).

#phase2
This folder contains the predicted bounding box for each frame in the ten videos. 

#phase3
This folder contains two folders: (i) xml_gt which is the ground-truth XML data for each video, and (ii) xml_predicted which contains the predicted results.

#trained_model
contains the trained Faster R-CNN model that we used for locating the selected file in a given image/frame.

#source_code
contains two folders: (i) VID2XML which contains the entire implementation of VID2XML, and (ii) replication which contains the code we used for replicating codeMotion and codeTube papers. 

#annotated_selected_file
contains the annotated selected file for the first fold