Songtao He   Email: songtao@alum.mit.edu songtaohe.com Resume GitHub Google Scholar Linkedin I received my Ph.D from the Department of Electrical Engineering and Computer Science (EECS) at MIT. My committee members are Hari Balakrishnan (Chair), Mohammad Alizadeh and Samuel Madden. My research interests are in computer systems and applied machine learning. More specifically, I'm interested in mobile computing and computer vision techniques that can help us create all kinds of digital maps about our world in a scalable and low-cost way. For example, I developed a series of automated techniques for building better digital street maps with aerial imagery and GPS data (see details). I'm also interested in building systems that can help improve developers' productivity in the face of complexity. For example, I developed BeeCluster, a drone orchestration platform that allows developers to easily program a fleet of drones, making drone programming more accessible to everyone.

Here are my recent projects,

• Enriching digital maps: Sat2Graph, RoadTagger, RoadRunner, Maid, RoadTracer, lane-level map extraction and crash risk mapping.
• Drone Programming Systems: BeeCluster (imperative) and SkyQuery (declarative).
• Faster inference and dataset labelling: MIRIS, TagMe, Unsupervised MOT, and SpatialFlow (a tool to speed up CNN inference on geospatial imagery).

---

Selected Publications (full list)

Lane-Level Street Map Extraction from Aerial Imagery Songtao He, Hari Balakrishnan Paper GitHub Presentation WACV, WAIKOLOA, Hawaii, Jan 2022.
Unsupervised Multi-Object Tracking with Dual-Tracker Consistency Favyen Bastani, Songtao He, Sam Madden Paper GitHub Presentation Web NeurIPS, virtual, Dec 2021.
Inferring high-resolution traffic accident risk maps based on satellite imagery and GPS trajectories Songtao He, Mohammad Amin Sadeghi, Sanjay Chawla, Mohammad Alizadeh, Hari Balakrishnan, Samuel Madden Paper GitHub Presentation ICCV, virtual, Oct 2021.
Sat2Graph: Road Graph Extraction through Graph-Tensor Encoding Songtao He, Favyen Bastani, Satvat Jagwani, Mohammad Alizadeh, Hari Balakrishnan, Sanjay Chawla, Mohamed M. Elshrif, Samuel Madden, Amin Sadeghi Paper GitHub Live Demo Presentation ECCV, Glasgow, Scotland, August 2020. Being used in production in ESRI ArcGIS. Number of Downloads: Loading
BeeCluster: Drone Orchestration via Predictive Optimization Songtao He, Favyen Bastani, Arjun Balasingam, Karthik Gopalakrishnan, Ziwen Jiang, Mohammad Alizadeh, Hari Balakrishnan, Michael Cafarella, Tim Kraska, Sam Madden Paper GitHub Web Presentation Slides MobiSys, Toronto, Canada, June 2020.
MIRIS: Fast Object Track Queries in Video Favyen Bastani, Songtao He, Arjun Balasingam, Karthik Gopalakrishnan, Mohammad Alizadeh, Hari Balakrishnan, Michael Cafarella, Tim Kraska, Sam Madden Paper GitHub Web SIGMOD, Portland, OR, June 2020.
RoadTagger: Robust Road Attribute Inference with Graph Neural Networks Songtao He, Favyen Bastani, Satvat Jagwani, Edward Park, Sofiane Abbar, Mohammad Alizadeh, Hari Balakrishnan, Sanjay Chawla, Samuel Madden, Mohammad Amin Sadeghi Paper GitHub AAAI, New York, NY, February 2020.
RoadRunner: improving the precision of road network inference from GPS trajectories Songtao He, Favyen Bastani, Sofiane Abbar, Mohammad Alizadeh, Hari Balakrishnan, Sanjay Chawla, Sam Madden Paper GitHub Web Slides ACM SIGSPATIAL, Seattle, WA, November 2018.
RoadTracer: Automatic Extraction of Road Networks from Aerial Images Favyen Bastani, Songtao He, Mohammad Alizadeh, Hari Balakrishnan, Samuel Madden, Sanjay Chawla, Sofiane Abbar, David DeWitt Paper GitHub Web CVPR, Salt Lake City, UT, June 2018
Machine-Assisted Map Editing Favyen Bastani, Songtao He, Sofiane Abbar, Mohammad Alizadeh, Hari Balakrishnan, Sanjay Chawla, Sam Madden Paper GitHub Web ACM SIGSPATIAL, Seattle, WA, November 2018.
Reducing Latency by Eliminating Synchrony Min Hong Yun, Songtao He, Lin Zhong Paper Web WWW, Perth, Australia, April 2017.
Optimizing Smartphone Power Consumption through Dynamic Resolution Scaling Songtao He, Yunxin Liu, Hucheng Zhou Paper Video ACM MobiCom, Paris, France, Sept 2015.

In Submission

SpatialFlow: Accelerating CNN Inference Tasks on Geo-Spatial Imagery with Irregular Regions of Interest Songtao He, Hari Balakrishnan, Mohammad Alizadeh
TagMe: GPS-Assisted Automatic Object Annotation in Videos Songtao He, Favyen Bastani, Mohammad Alizadeh, Hari Balakrishnan, Michael Cafarella, Tim Kraska, Samuel Madden Paper

Experience

• Research Intern, Microsoft Research Redmond, Mentored by Dr. Sanjeev Mehrotra, Summer 2019
• Visiting, Rice Efficient Computing Group, Mentored by Prof. Lin Zhong, 2015 to 2016
• Research Intern, Microsoft Research Asia, Mentored by Dr. Yunxin Liu, 2014 to 2015
• Research Intern, University of Hong Kong, Mentored by Prof. Cho-li Wang, Summer 2014

Teaching

• TA, 6.858 Computer System Security, Spring 2022
• TA, 6.S062 Mobile and Sensor Computing, Spring 2017

Professional Service

• Reviewer of CVPR (2021, 2022), ICCV (2021), AAAI (2021, 2022), ICLR (2022), NeurIPS (2021), WACV (2022), SDM (2022), and ECCV (2022)
• External Reviewer of ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD) 2021
• Reviewer of IEEE Transactions on Neural Networks and Learning Systems, 2021
• Reviewer of ACM Transactions on Spatial Algorithms and Systems, 2020
• Reviewer of IEEE Transactions on Pattern Analysis and Machine Intelligence, 2019
• Reviewer of IEEE Geoscience and Remote Sensing Letters, 2019

Talk

• BeeCluster: Drone Orchestration via Predictive Optimization
  ACM MobiSys, June 2020
• The Science of Algorithmic Map Inference (Tutorial at KDD'18)
  ACM SIGKDD, London, UK, August 2018