Zeeshan Zia

Zeeshan Zia
Redmond, WA 98052
zeeshan{@}retrocausal{.}ai

CEO and Co-Founder
Retrocausal Inc.

Retrocausal Logo

We are on a journey to build a truly scalable Manufacturing-as-a-Service platform leveraging AI. We begin with live worker guidance and prescriptive analytics to make existing manufacturing more efficient.

I have authored more than 15 peer-reviewed publications on visual activity understanding, 3D scene understanding, learning from synthetic data, and machine learning systems, which have been cited by every major lab from MIT, Stanford, and CMU, to Apple, Google, and Baidu.

I am also an inventor on more than a dozen US patents or patent applications, and have shipped AI-first products including Microsoft HoloLens, driver-assistance on Honda cars, and the first AR-capable Lego toys.

Curriculum Vitae

External Links

Industry

Microsoft Research
Redmond, WA
...
Senior Scientist (HoloLens)
2017-2019

NEC Laboratories America
Cupertino, CA
...
Researcher
2015-2017

Qualcomm Research
Vienna, Austria
...
Research Intern
Summer 2013

Siemens Corp. Technologies
Munich, Germany
...
Engineering Intern
Summer 2008

Academia

Postdoc
Imperial College London
...
London, UK
2014-2015

PhD
Swiss Federal Institute of Technology
...
Zurich, Switzerland
2009-2013

MS
Munich University of Technology
...
Munich, Germany
2007-2009

Selected Recent Publications see all...

  • S. Haresh, S. Kumar, M.Z. Zia, Q.H. Tran. Towards Anomaly Detection in Dashcam Videos.31st IEEE Intelligent Vehicles Symposium 2020Conference
    Inexpensive sensing and computation, as well as insurance innovations, have made smart dashboard cameras ubiquitous. Increasingly, simple model-driven computer vision algorithms focused on lane departures or safe following distances are finding their way into these devices. Unfortunately, the long-tailed distribution of road hazards means that these hand-crafted pipelines are inadequate for driver safety systems. We propose to apply data-driven anomaly detection ideas from deep learning to dashcam videos, which hold the promise of bridging this gap. Unfortunately, there exists almost no literature applying anomaly understanding to moving cameras, and correspondingly there is also a lack of relevant datasets. To counter this issue, we present a large and diverse dataset of truck dashcam videos, namely RetroTrucks, that includes normal and anomalous driving scenes. We apply:(i) one-class classification loss and (ii) reconstruction-based loss, for anomaly detection on RetroTrucks as well as on existing static-camera datasets. We introduce formulations for modeling object interactions in this context as priors. Our experiments indicate that our dataset is indeed more challenging than standard anomaly detection datasets, and previous anomaly detection methods do not perform well here out-of-the-box. In addition, we share insights into the behavior of these two important families of anomaly detection approaches on dashcam data.
    @inproceedings{haresh20iv,
     author = {S. Haresh and S. Kumar and M.Z. Zia and Q.H. Tran},
     title = {Towards Anomaly Detection in Dashcam Videos},
     booktitle = {31st IEEE Intelligent Vehicles Symposium (IV)},
     year = {2020}
    }
  • H. Coskun, M.Z. Zia, B. Tekin, F. Bogo, N. Navab, F. Tombari, H. Sawhney. Domain-Specific Priors and Meta Learning for Low-shot First-Person Action Recognition. arXiv 2019Technical Report
    The lack of large-scale real datasets with annotations makes transfer learning a necessity for video activity understanding. Within this scope, we aim at developing an effective method for low-shot transfer learning for first-person action classification. We leverage independently trained local visual cues to learn representations that can be transferred from a source domain providing primitive action labels to a target domain with only a handful of examples. Such visual cues include object-object interactions, hand grasps and motion within regions that are a function of handlocations. We suggest a framework based on meta-learning to appropriately extract the distinctive and domain invariant components of the deployed visual cues, so to be able to transfer action classification models across public datasets captured with different scene configurations. We thoroughly evaluate our methodology and report promising results over state-of-the-art action classification approaches for both inter-class and inter-dataset transfer.
    @inproceedings{coskun19arxiv,
     author = {H. Coskun and M.Z. Zia and B. Tekin and F. Bogo and N. Navab and F. Tombari and H. Sawhney},
     title = {Domain-Specific Priors and Meta Learning for Low-shot First-Person Action Recognition},
     booktitle = {arXiv:1907.09382},
     year = {2019}
    }