Samia Shafique
Thank you for taking an interest in my profile. I am a final-year Ph.D. candidate in the Department of Computer Science at the University of California Irvine (UCI). My research area is Computer Vision and I work on projects encompassing multimodal image retrieval, depth estimation, intrinsic image decomposition, and domain adaptation under the supervision of Professor Charless Fowlkes.
Take a look at my recent work published in WACV 2023!
sshafiqu AT uci DOT edu
Research
S Shafique, S Kong, CC Fowlkes, "CriSp: Leveraging Tread Depth Maps for Enhanced Crime-Scene Shoeprint Matching", in the 18th European Conference on Computer Vision (ECCV) 2024.
Investigating crime-scene shoeprints to identify the make and model of the shoe that left a print can help narrow down suspects of a crime. We
• Introduce concept of matching crime-scene shoeprints to tread depth maps, aiming to facilitate forensic investigation and criminal justice.
• Propose a new benchmark consisting of a new dataset and retrieval-based evaluation protocols, allowing fair comparisons against previous methods.
• Learn feature representation from cross-domain, noisy, and occluded crime-scene prints
• Develop a spatially-aware matching method named CriSp, which demonstrates superior performance over existing methods in both shoeprint matching and image retrieval tailored to this specific task.
S Shafique, B Kong, S Kong, CC Fowlkes, "Creating a Forensic Database of Shoeprints from Online Shoe Tread Photos", in IEEE Winter Conference on Applications of Computer Vision (WACV) 2023.
To aid the study of crime-scene shoeprints, we propose to create a database of footwear impressions that cover the large and growing number of distinct shoe models by leveraging shoe-tread photos collected by online retailers. For this project, we
• Render a synthetic dataset of shoe-treads with ground-truth intrinsics from shoeprint images
• Design and train a CNN for image decomposition into albedo, depth, normal, and light
• Learn from unlabeled real data using intrinsic image decomposition and domain adaptation
• Outperform relevant state-of-the-art methods (e.g., RIN, CyCADA)
• Build a large-scale database of shoe-treads with RGB photo, depth, and shoeprint information
Albedo Segmentation for Single Objects
Estimate the reflectance given single images. We find that many everyday objects (such as shoes, dresses, etc) consist of only a handful of colors. In this project, we segment images of such objects using a variation of the Mean Shift algorithm to obtain a reflectance map.
Pin Sequence Generation for String Art
Inspired by Petros Vrellis, we produce threaded recreations of images. The core technical challenge is to find a pin sequence through which to wound a thread given a loom with pins along its circumference.
S Shafique, ME Ali, "Recommending Most Popular Travel Path within a Region of Interest from Historical Trajectory Data", ACM SIGSPATIAL International Workshop on Mobile Geographic Information Systems, MobiGIS, 2016.
We propose a method to extract the most popular travel path within a Region of Interest (ROI) from historical Flicker photo uploads.
• Challenge: user trajectories are inaccurate and incomplete
• Method: rephrase trajectories into smaller parts and eliminate noisy points
Posters and Publications
Publications
Samia Shafique, Bailey Kong, Shu Kong, Charless C. Fowlkes, "Creating a Forensic Database of Shoeprints from Online Shoe Tread Photos", in IEEE Winter Conference on Applications of Computer Vision (WACV) 2023 Jan 3. [project page, dataset, code, pdf, poster, video]
Samia Shafique, Muhammad Eunus Ali, "Recommending Most Popular Travel Path within a Region of Interest from Historical Trajectory Data", In Proceedings of the Fifth ACM SIGSPATIAL International Workshop on Mobile Geographic Information Systems, MobiGIS ’16, San Francisco, CA, USA, October 31, 2016. [pdf, poster, video]
Poster Presentations
Samia Shafique, Muhammad Eunus Ali, "Recommending the Most Popular Travel Path Within a Region", selected for Top 11 Poster Awards in Undergraduate Thesis Poster Presentation, CSE, BUET, Dhaka, Bangladesh, February 2016. [pdf, poster, video]
Samia Shafique, Muhammad Eunus Ali, "A Platform to Enhance Online Shopping Experience", Poster presented at Broadening Participation Workshop, Ubicomp, Osaka, Japan on September 8th, 2015. [pdf, poster]
Syeda Farzia Afroze, Samia Shafique, Farzana Ahmed Siddique, "Enhancing Classroom Communication and Academic Task Management through Mobile Devices", Poster presented at the Grace Hopper Celebration India on November 20th -22nd 2014. [pdf, poster]
Samia Shafique, Zaheen Mustakin Ayon, "Transportation: Tracking and Preventing Accidents", Poster presented at Grace Hopper Celebration, Phoenix, Arizona on October 8th - 10th, 2014. (acceptance rate 22%) [pdf, poster]
Work Experience
June 2019 - Present
As a Graduate Research Assistant (GSR), I work with Professor Charless Fowlkes at the Computer Vision lab at UC Irvine. My research work focuses on aiding investigators with crime-scene shoeprint analysis and encompasses depth prediction, domain adaptation, and multimodal image retrieval.
I have created a large-scale synthetic database of shoes with detailed tread information by predicting depth from shoe-tread images downloaded from online retailers. The make and model of crime-scene shoeprints are identified by matching the prints to our curated database and retrieving the most similar shoes.
For these projects, I have rendered a large-scale synthetic dataset and collected real datasets for training and evaluation. I have also analyzed different state-of-the-art networks extensively and proposed new methods that outperform them.
Projects I have worked on as a GSR are listed here.
June 2023 - September 2023
I studied document understanding to automate fraud detection. Here are a few things I have gained experience with during the course of the internship:
Prompt engineering for Chat GPT
Fine-tuning an LLM (T5) and working with hugging face models
Experimenting with custom prompt and response formats for LLMs
September 2019 - September 2021
The Diverse Educational Community and Doctoral Experience (DECADE) aims to create a warm and inclusive climate for students at the University of California, Irvine. As the Student Representative of the Donald Bren School of Information and Computer Sciences (ICS) at UC Irvine, I helped in organizing events and initiatives involving faculty for a supportive environment for all graduate students.
September 2017 - June 2019
I was responsible for both Bachelor's and Master's level courses including Computer Vision, Programming, and Computational Photography. As a Teaching Assistant (TA) I assisted students with course objectives, conducted exams, and graded scripts among other things. A full list of courses I was TA for is available here.
June 2018 - September 2018
I worked on various aspects of the Polydyne software (a product of Supplyframe) including designing and implementing 3 features from scratch. The features are a generic module for exporting data, a generic progress display page, and a scheduled background task to clean up old files from the cloud storage used by the software. In addition to these, I have worked on bug fixes and minor improvements in the software.
May 2016 - May 2017
As a lecturer in the Computer Science department, my primary task was to design and instruct theory and laboratory courses on Operating Systems, Graphics, Algorithms, and Digital Logic Design. In addition to this, I have volunteered to be a Judge for projects at Project Showcasing, UIU CSE Festival 2017, and an Instructor and Problem Setter at ICT Olympiad and UIU Programming Camp 2016. A list of my activities is available here.
Vice President, June 2014 - February 2016
Event Director, May 2014 - May 2015
I am a Founding Member of Bangladeshi Women in Computer Science in Engineering. While serving as a Vice President and Event Director there, I organized events to encourage women toward higher studies and career pursuit in STEM. My activities included motivating students to participate in academic research, competitions, development work, etc. It was also my immense pleasure to initiate a formal 3-tier mentoring program to help undergraduate students learn from experienced seniors and alumni.
Awards and Scholarships
Awards
Top 11 Posters Award, Thesis Poster Presentation at CSE, BUET, 2016
Outstanding Performance in a Leadership Role, BWCSE, 2016 [link]
International Champion and ITPEC Gold Award Winner at ITEE FE exam organized by BDITEC 2015 [link]
Champion at BWIT Mobile Apps Contest, 2015
National Champion at International Robotics Challenge (IRC), International position – 9th, 2014 [link]
18th at EESTECs Android Competition, 2014
Top 2 Projects Award at Firefox Open Source Hackathon, 2014
11th at the Firefox OS App Day, 2014
Champion at Robotics, CSE festival, 2013
Scholarships
Travel scholarship recipient for Grace Hopper Celebration, October 11th - 14h 2019, held at Orlando, Florida, USA
Dean's Top Off Award 2017, University of California, Irvine
Google Anita Borg Memorial Scholarship Asia Pacific 2015 (43/1000+) [link]
Scholarship recipient for Experiencing HPC for Undergrads, SC15, November 15th - 20th 2015, held at Austin, Texas
Scholarship recipient at Broadening Participation Workshop, Ubicomp, September 8th 2015, held at Osaka, Japan
Scholarship recipient at Grace Hopper Celebration, October 8th - 10th 2014, held at Phoenix, Arizona, USA (acceptance rate 26%)
Scholarship recipient at Grace Hopper Celebration India, November 19th - 21st 2014, held at Bangalore, India
BUET Dean's Award
BUET University Merit and Stipend Scholarship
Artwork
Here's a sneak peak into my world of art! Visit my album for more.