PeopleofSuperintelligence

Researcher

Menglong Zhu

Menglong Zhu is a computer vision engineer and a coauthor of MobileNets, the compact neural networks Google released in 2017 for on-device and embedded vision. Trained at Fudan University and the University of Pennsylvania, they completed a PhD at Penn's GRASP Lab under Kostas Daniilidis in 2016. At Google, Zhu was a core team member of the entry that won the 2016 COCO detection challenge, contributed to the TensorFlow Object Detection API, and coauthored MobileNetV2 and an integer-only quantization scheme for mobile inference. They now work at DJI, where they coauthored a 2022 binary neural network framework for on-device computer vision.

Now at
DJI
Works
42
Citations
58,192
h-index
19
Badges

Publication numbers from OpenAlex, as of 22 October 2026.

Timeline

  1. 2010–2012

    Fudan and a Penn master's

    Zhu earns a bachelor's degree in computer science at Fudan University in 2010. In 2012, they complete a master's degree in robotics at the University of Pennsylvania.

  2. 2013–2014

    Actemes and deformable part models

    In 2013, they coauthor the ICCV paper "From Actemes to Action" with Weiyu Zhang and Konstantinos G. Derpanis, introducing a video dataset that labels the action and body keypoints in every frame. At ECCV in 2014, they publish Active Deformable Part Models Inference, a fast object detector, with Nikolay Atanasov, George J. Pappas and Kostas Daniilidis.

  3. 2016

    A GRASP Lab PhD

    In 2016, they complete a PhD in computer information science at Penn's GRASP Lab, advised by Kostas Daniilidis, and then join Google. With Jonathan Huang, Kevin Murphy and other researchers, they coauthor a November 2016 guide to speed and accuracy trade-offs in Faster R-CNN, R-FCN and SSD object detectors.

  4. 2016–2017

    COCO win and the detection API

    Zhu is a core team member of Google's G-RMI entry, which places first in the 2016 COCO detection challenge. In June 2017, Google releases that detection system as the TensorFlow Object Detection API, an open-source framework for building, training and deploying detection models, with Zhu as one of its core contributors.

  5. 2017

    MobileNets for on-device vision

    In April 2017, they coauthor MobileNets with Andrew G. Howard, Bo Chen, Dmitry Kalenichenko and others; the models use depthwise separable convolutions and two global hyperparameters that trade off latency and accuracy. Google releases MobileNets for TensorFlow in June 2017, announced in a blog post by Howard and Zhu.

  6. December 2017–January 2018

    Integer inference and MobileNetV2

    In December 2017, Zhu coauthors a quantization scheme that runs neural network inference with integer-only arithmetic, together with a training procedure that preserves accuracy after quantization. MobileNetV2, written with Mark Sandler, Andrew Howard, Andrey Zhmoginov and Liang-Chieh Chen, appears in January 2018; it is built on inverted residual blocks with thin bottleneck layers and adds the SSDLite detection framework.

  7. March 2019

    Memory-guided mobile video detection

    In March 2019, they coauthor "Looking Fast and Slow" with Mason Liu, Marie White, Yinxiao Li and Dmitry Kalenichenko. The video object detector pairs lightweight feature extractors with a temporal memory and runs at speeds of up to 70+ FPS on a Pixel 3 phone.

  8. As of October 2026

    Binary networks at DJI

    In February 2022, with colleagues at DJI Innovations, Zhu coauthors a binary neural network framework that overtakes 8-bit quantization in the speed-vs-accuracy tradeoff for classification, detection, segmentation, super-resolution and matching. They work at DJI.

Career

  1. Since 2022DJI
  2. 2023–2025Huai'an University
  3. 2016–2020Google

Education

  1. 2016University of PennsylvaniaPhD in computer information science
  2. 2012University of PennsylvaniaMaster's degree in robotics
  3. 2010Fudan UniversityBachelor's degree in computer science

From the biography and OpenAlex affiliations (the years of papers that list the organization), so some years are approximate.

Most-cited works

  1. MobileNetV2: Inverted Residuals and Linear Bottlenecks2018 · IEEE Computer Society Conference on Computer Vision and Pattern Recognition · 28,599 citationsWith Andrew Howard and Mark Sandler
  2. MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications2017 · 19,002 citationsWith Andrew Howard and Hartwig Adam
  3. Quantization and Training of Neural Networks for Efficient Integer-Arithmetic-Only Inference2018 · IEEE Computer Society Conference on Computer Vision and Pattern Recognition · 4,068 citationsWith Andrew Howard and Hartwig Adam
  4. Speed/Accuracy Trade-Offs for Modern Convolutional Object Detectors2017 · IEEE Computer Society Conference on Computer Vision and Pattern Recognition · 2,740 citationsWith Kevin P. Murphy and Sergio Guadarrama
  5. Rethinking RGB-D Salient Object Detection: Models, Data Sets, and Large-Scale Benchmarks2020 · IEEE Transactions on Neural Networks and Learning Systems · 710 citationsWith Ming‐Ming Cheng

From OpenAlex, as of 21 October 2026.

  1. Google had the most people shaping AI every year from 2014 to 2025Includes Google. Our estimates of which organizations employed the most people shaping AI each year from 1955 to 2025, tracing the shift from university labs to companies.
  2. Google has the highest combined h-index among AI organizationsIncludes Google. We estimate that Google holds the highest combined h-index at 8,016, while universities account for 73% of the deduplicated total across AI organizations.
  3. 20% of people at companies who shape AI also hold a university role, led by Google and Stanford UniversityIncludes Google. We estimate that 539 of the 6,766 people with a known current organization (8%) hold both a university role and a company role, including 20% of people at companies and 13% of people at universities.
  4. Google alumni have founded or led 133 other organizations, more than any other AI labIncludes Google. We estimate that alumni of 17 major AI labs went on to found or lead 350 other organizations, including 226 they founded, and we trace 472 career moves directly between 23 labs since 2010.

Colleagues

People named in this biography, and people with dated positions at the same organization in the same years.

Sources and credits

Photo: We have no published photo of Menglong Zhu yet. Send us a free-licensed photo.

  1. Menglong Zhu's Homepage - Computer Vision & Robotics dreamdragon.github.io. Checked 20 October 2026.
  2. ICCV 2013 Open Access Repository: From Actemes to Action: A Strongly-Supervised Representation for Detailed Action Understanding openaccess.thecvf.com. Checked 20 October 2026.
  3. [1611.10012] Speed/accuracy trade-offs for modern convolutional object detectors arxiv.org. Checked 20 October 2026.
  4. G-RMI Object Detection research.google. Checked 20 October 2026.
  5. Supercharge your Computer Vision models with the TensorFlow Object Detection API research.google. Checked 20 October 2026.
  6. [1704.04861] MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications arxiv.org. Checked 20 October 2026.
  7. MobileNets: Open-Source Models for Efficient On-Device Vision research.google. Checked 20 October 2026.
  8. [1712.05877] Quantization and Training of Neural Networks for Efficient Integer-Arithmetic-Only Inference arxiv.org. Checked 20 October 2026.
  9. [1801.04381] MobileNetV2: Inverted Residuals and Linear Bottlenecks arxiv.org. Checked 20 October 2026.
  10. [1903.10172] Looking Fast and Slow: Memory-Guided Mobile Video Object Detection arxiv.org. Checked 20 October 2026.
  11. [2202.03716] Binary Neural Networks as a general-propose compute paradigm for on-device computer vision arxiv.org. Checked 20 October 2026.
  12. Kostas Daniilidis - computer vision and robotics cis.upenn.edu. Checked 20 October 2026.
  13. Menglong Zhu – GRASP Lab grasp.upenn.edu. Checked 20 October 2026.