Unsupervised Discovery of Character Dictionaries in Animation Movies
Author: Krishna Somandepalli
Project Wiki: https://github.com/usc-sail/mica-animation/wiki
Affilitation:
Signal Analysis and Interpretation Laboratory (SAIL)
University Southern California (USC)
Los Angeles, CA, USA
Contact: somandep@usc.edu
This repository contains data relevant for publication titled:
"Unsupervised Discovery of Character Dictionaries in Animation Movies"
Krishna Somandepalli, Naveen Kumar, Tanaya Guha, Shrikanth Naryanan
SAIL, USC, Los Angeles, CA, USA
As of July 17 2017, this paper was accepted for publication in the IEEE Transactions in Multimedia pending modifications
Please refer to the paper to understand the different steps in the methodology and the referred scripts The scripts are documented as per Figure 3 (Overview schematic diagram) image in the paper
Scripts:
- MultiBox DNN Object Detector:
-- video_deep_multibox_detect.py - Coarse detection of character candidates:
-- choose_object_candidates.py - Saliency Constraints
-- choose_object_candidates_SAL.py - Local Tracking in video
-- easy_video_object_local_tracking.py - Clustering character candidates
-- experimenter_cluster_candidates.py - Peformance Analysis Measures
-- performance_analysis_measures.py
Other directories:
- Mechanical Turk (MTurk) annotations and parsing scripts in directory:
-- mturk_annotations/ - Clustering results in:
-- experimental_results/
Data description and annotation labels:
First, fill out this Google form with the required information to receive the password.
Then, use the password to download the data at this link for the outputs and annotations used in the system evaluation of the paper