Qualcomm Inc.

10/25/2021 | News release | Distributed by Public on 10/26/2021 21:26

Innovative solutions from the 2021 AI Developer Conference

The Qualcomm AI Developer Conference was held in September 2021 in Chengdu, China, and the theme this year was presenting a new era of the intelligent interconnection of technologies. Heng Xu, Sr. Dir of Product Marketing from Qualcomm International (Shanghai) Co., Ltd., presented our AI strategy and new technologies. While Song Mao, Sr. Dir of Ventures from Qualcomm Wireless Communication Technologies (China) Limited, shared additional AI investment plans in China.

The conference saw over 360+ attendees with speakers from Qualcomm Technologies, Inc., ThunderSoft, Xiaomi, Extreme Vision, Jiijan Tech, and FutureMind. There was a wide range of topics including, AI product technology upgrades, Artificial Intelligence of Things (AIoT), on device AI development trends, and mobile applications.

A key highlight from the conference was the awards ceremony to announce the 2021 Qualcomm AI Innovation Challenge Award winners, which saw participation from over 1,300 developers. The contest provided developers with end-to-end AI development tools and support to help them create high-quality, low-cost, commercialized-grade AI applications more efficiently.

The tools included the Qualcomm Neural Processing SDK and the Qualcomm AI Engine Direct architecture. The architecture, a full-stack AI solution for our latest-generation Snapdragon mobile platforms, provides developers access to the SoC Qualcomm Hexagon DSP, Qualcomm Adreno GPU, and Qualcomm Kryo CPU.

There were sixteen winners chosen at the award ceremony, and here is a brief overview of the top four:

  • Shining Cloud (Shenzhen) Technology Co., Ltd. developed their High performance under display ToF sensing with AI, allowing mobile phones to perform features like eye tracking, facial recognition, heartbeat detection and gesture recognition using time-of-flight (ToF) technology. This allows ToF sensors to be directly linked to applications making it more efficient to calculate complex signals, to generate infrared and depth data. The project utilizes Hexagon hardware acceleration to improve real-time performance, stability, and inference efficiency.
  • Shenzhen CMCross Technology Co., Ltd. developed their Driver Biometrics System (aka Non-contact driver's health perception system). This application captures images through an in-vehicle camera and uses a unique image processing algorithm to detect physiological indicators of drivers and passengers. Car manufacturers can then use the information for human-machine coupling for improving driving experience and safety.
  • Precision Computing Team lead by Li Yang, developed "Abstract Hook Xuan", an intelligent video summarization system at the edge, backed by cloud-side AI. Using deep learning-based video summarization and text-summarization technologies, the solution captures a user-submitted video link, and watches that video on their behalf to provide information about it. For example, with a live broadcast, the solution can generate graphic abstracts containing the key content of the video, to help users find graphic content quickly and efficiently.
  • Fourth winner (AIOT award): Segway-Ninebot developed the T60 self-driving scooter. Their computer vision project performs object detection and segmentation (e.g., to identify, pedestrians, road features, etc.) to help navigate the scooter. The team listed numerous use cases including, detection of dense pedestrian traffic to automatically limit the scooter's top speed, and remotely driving of scooters (e.g., by scooter rental companies) where the unit automatically handles obstacle avoidance.

These entries materialized through the hard work of each participant, and backed by AI hardware and software solutions from Qualcomm Technologies, Inc. This year's challenge saw many amazing AI applications competing across a range of categories and showcasing how AI can solve many different problems. While these represent just a small sample of the 16 winners and the efforts of over 1,300 developers, it's clear to see just how AI applications can touch many aspects of our lives. We can't wait to see what future innovation challenges will bring!

Ready to incorporate AI into your next project? Be sure to check out the following resources. And if you've built something cool using any of our technologies, be sure to tell us about it, and we may just feature it on Qualcomm Developer Network.

Qualcomm tools and resources:

Recent blog posts:

Snapdragon, Qualcomm Neural Processing SDK, Qualcomm AI Engine, Qualcomm Hexagon, Qualcomm Adreno, and Qualcomm Kryo are products of Qualcomm Technologies, Inc. and/or its subsidiaries. AIMET is a product of Qualcomm Innovation Center, Inc.