Jump to content

DeepScale

From Wikipedia, the free encyclopedia
This is an old revision of this page, as edited by DGG (talk | contribs) at 17:05, 15 October 2019 (copyed, removing links that do not really add information.). The present address (URL) is a permanent link to this revision, which may differ significantly from the current revision.
DeepScale, Inc.
Type of site
Privately held company
HeadquartersMountain View, CA
Founder(s)
CEOForrest Iandola
URLhttp://deepscale.ai

DeepScale, Inc. is a privately held, US-based technology company headquartered in Mountain View, CA,that develops perceptual system technologies for automated vehicles. On October 1st, 2019 it was purchased by Tesla.[1]

History

DeepScale was co-founded by Dr. Forrest Iandola and Prof. Kurt Keutzer in September 2015., [2] who had worked together at University of California, Berkeley on neural networks. .[3][4] In 2018, DeepScale raised US $15 Million in Series A funding.[5] In 2018, the firm announced strategic partnerships with automotive suppliers including Visteon and Hella Aglaia Mobile Vision GmbH.[6][7] On October 1st, 2019 the firm was purchased by Tesla, the leading producer of autonomous cars. [1]

Technology

Prior to the founding of DeepScale, Forrest Iandola and Kurt Keutzer worked together at University of California, Berkeley, on making deep neural networks (DNNs) more efficient.[3][4] In 2016, shortly after the founding of DeepScale, Iandola, Keutzer, and their collaborators released SqueezeNet, which is a small and energy-efficient DNN.[8][9] By developing smaller DNNs, DeepScale has been able to run deep learning on scaled-down processing hardware such as smartphones and automotive-grade chips.[10][8][11] In 2018, DeepScale said that its engineering team had moved beyond SqueezeNet and that it had developed even faster and more accurate DNNs for use in commercial products.[12]

In recent years, Neural architecture search (NAS) has begun to outperform humans at designing DNNs that produce high-accuracy results while running fast.[13] In 2019, DeepScale published a paper called SqueezeNAS, which used supernetwork-based NAS to design a family of fast and accurate DNNs for semantic segmentation of images.[14] DeepScale's paper claimed that the SqueezeNAS neural networks outperform the speed-accuracy tradeoff curve of Google's MobileNetV3 family of neural network models.[15] Further, while Google used thousands of GPU-days to search for the design of MobileNetV3, DeepScale used just tens of GPU-days to automatically design the DNNs presented in the SqueezeNAS paper.[16]

Product

DeepScale develops perceptual system software which uses deep neural networks to enable cars to interpret their environment. The software is designed for integration into an open platform, where a wide range of sensors and processors can be used.[6] The software is able to run on a variety of processors, ranging from NVIDIA GPUs to smaller ARM-based processing chips that are designed specifically for the automotive market.[6][12]

In January 2019, the firm launched an automotive perception software product, Carver, that uses deep neural networks to perform object detection, lane identification, and drivable area identification. To accomplish this, Carver uses three neural networks which run in parallel. While running in real-time, these three networks perform a total of 0.6 tera-operations per second.[17] As a point of reference, 0.6 tera-ops/sec is only 2 percent of the 30 tera-ops/sec that the NVIDIA Jetson Xavier embedded computing system is rated to perform.[18]

Awards and recognition

In December 2016, VentureBeat named DeepScale one of the "15 interesting startups to watch in 2017." [19] In December 2018, the AI Time Journal listed DeepScale as one of the "Top 25 Artificial Intelligence Companies of 2018."[20][21] In February 2019, DeepScale was included in the CB Insights AI 100, which consists of "the most promising 100 AI startups working across the artificial intelligence value chain."[22]

References

  1. ^ a b Kolodny, Lora (2019-10-01). "Tesla is buying computer vision start-up DeepScale in a quest to create truly driverless cars". CNBC. Retrieved 2019-10-02.
  2. ^ "DeepScale". Crunchbase. Retrieved 2018-04-07.
  3. ^ a b Keutzer, Kurt. "Faculty Webpage". UC Berkeley. Retrieved 2018-05-22.
  4. ^ a b Keutzer, Kurt. "Students". UC Berkeley. Retrieved 2018-05-22.
  5. ^ Marinova, Polina (2018-04-04). "Term Sheet". Fortune. Retrieved 2018-05-22.
  6. ^ a b c Yoshida, Junko (2018-01-09). "Visteon Works with DNN Vanguard DeepScale". EE Times. Retrieved 2018-04-07.
  7. ^ Yoshida, Junko (2018-04-03). "Are We Short of Deep Learning Experts?". EE Times. Retrieved 2018-04-07.
  8. ^ a b Yoshida, Junko (2017-09-21). "DeepScale on Robo-Car: Fuse Raw Data". EE Times. Retrieved 2018-05-22.
  9. ^ Iandola, Forrest N; Han, Song; Moskewicz, Matthew W; Ashraf, Khalid; Dally, William J; Keutzer, Kurt (2016). "SqueezeNet: AlexNet-level accuracy with 50x fewer parameters and <0.5MB model size". arXiv:1602.07360 [cs.CV].
  10. ^ Cite error: The named reference :1 was invoked but never defined (see the help page).
  11. ^ Shazar, Jon (2018-04-05). "Steve Cohen Buys The Dip In Self-Driving Cars". Dealbreaker. Retrieved 2018-05-22.
  12. ^ a b "How to become a Full-Stack Deep Learning Engineer (time: 51:30)". Silicon Valley Deep Learning Group. Retrieved 2018-05-22.
  13. ^ Zoph, Barret; Vasudevan, Vijay; Shlens, Jonathon; Le, Quoc V. (2017-07-21). "Learning Transferable Architectures for Scalable Image Recognition". arXiv:1707.07012 [cs.CV].
  14. ^ Shaw, Albert; Hunter, Daniel; Iandola, Forrest; Sidhu, Sammy (2019). "SqueezeNAS: Fast neural architecture search for faster semantic segmentation". arXiv:1908.01748 [cs.CV].
  15. ^ Howard, Andrew; Sandler, Mark; Chu, Grace; Chen, Liang-Chieh; Chen, Bo; Tan, Mingxing; Wang, Weijun; Zhu, Yukun; Pang, Ruoming; Vasudevan, Vijay; Le, Quoc V.; Adam, Hartwig (2019-05-06). "Searching for MobileNetV3". arXiv:1905.02244 [cs.CV].
  16. ^ Yoshida, Junko (2019-08-25). "Does Your AI Chip Have Its Own DNN?". EE Times. Retrieved 2019-09-26.
  17. ^ Landen, Ben (2019-01-25). "DeepScale Announces Carver21: Modular Deep Learning Perception Software for Driver-Assistance". DeepScale Blog. Retrieved 2019-02-04.
  18. ^ NVIDIA, Corporation (2018-06-03). "Next-Gen Robotic Systems to Be Enabled by Jetson Xavier Computer and Isaac Robotics Software". NVIDIA. Retrieved 2019-02-04.
  19. ^ Yueng, Ken (2016-12-29). "15 interesting startups to watch in 2017". VentureBeat. Retrieved 2019-04-24.
  20. ^ "TOP 25 Artificial Intelligence Companies 2018". AI Time Journal. 2018-12-18. Retrieved 2019-06-06.
  21. ^ "Interview with Forrest Iandola, CEO and Co-Founder of DeepScale". AI Time Journal. 2019-03-18. Retrieved 2019-04-24.
  22. ^ "AI 100: The Artificial Intelligence Startups Redefining Industries". CB Insights. 2019-02-06. Retrieved 2019-04-24.
Retrieved from "https://en.wikipedia.org/w/index.php?title=DeepScale&oldid=921420562"