Sentry: Artificial Intelligence Solution for Cameras

Sentry (https://www.smarthomesentry.com/) provides Artificial Intelligence solution for the surveillance cameras. Sentry's solution is a plugable AI layer on top of existing IP enabled video surveillance cameras.
Sentry's cutting-edge Deep Learning solution enables:
a. Sending preventive alerts by using AI to detect possible threats.
b. Sending only meaningful alerts to users.
c. Rapid response to an unusual/threat event (with image/video link)
d. More importantly, affordable solution without needing to purchase the new equipment.
 
 

Outline/Structure of the Talk

a. Introduction to application of Deep Learning for Computer Vision

b. Building Threat detection systems using Video Camera Feeds

c. Scaling issues/challenges with large data feeds

d. What is unique about Sentry

e. Closing comments

f. Q&A

Learning Outcome

a. Applications of Deep Learning in Image Recognition & threat detection.

b. Challenges with Image Processing on Video Feeds from cameras.

c. How to architect, deploy and scale the Deep Learning systems for Computer Vision applications

Target Audience

Machine Learning and Computer Vision enthusiasts

Prerequisites for Attendees

Basics of Computer Vision and Deep Learning

schedule Submitted 7 months ago

Public Feedback

comment Suggest improvements to the Speaker
  • Dr. Vikas Agrawal
    By Dr. Vikas Agrawal  ~  7 months ago
    reply Reply

    Dear Keshav: Do you work for or lead Smart Home Sentry's work? I would love to learn more. From your profile it appears that you work for Infinera.

    Warm Regards

    Vikas

    • Keshava Murali Elliadka
      By Keshava Murali Elliadka  ~  7 months ago
      reply Reply

      Hello Vikas,

                           Thanks for reaching out.

      Yes. I work as consultant for Sentry. I have uploaded the video/slide deck with my intro included (apologies for not doing it earlier).

      Best,

      Keshava

      • Dr. Vikas Agrawal
        By Dr. Vikas Agrawal  ~  6 months ago
        reply Reply

        Thank you for sharing, Keshava. I have a couple of questions on your proposal:

        1. Do you have explicit permission from the ML team of your client Sentry or from Uday to share details of their algorithms or will you present primarily a general overview? As you know all the presentation materials and videos from ODSC are shared publicly.

        2. As you work on driver development primarily, do you specifically work together with the ML team to deliver specific aspects of the algorithms that will be presented?

        Warm Regards,

        Vikas

        • Keshava Murali Elliadka
          By Keshava Murali Elliadka  ~  6 months ago
          reply Reply

          Hello Vikas,

                           Please find my answers.

          1. My presentation would be a general overview of the Sentry solution. I will touch upon some of the algorithms we use, but may not be able to share the specific details.

          2. Yes. I work closely with the team in Sunnyvale to deliver and productize the algorithms and integrate them to product pipeline.

          Please let me know if you have any other questions. Thanks !

          Best,

          Keshava

          • Dr. Vikas Agrawal
            By Dr. Vikas Agrawal  ~  6 months ago
            reply Reply

            Dear Keshava: At ODSC the audience enjoys and relishes algorithmic details that are shared so that they can go back and apply at their own workplace. Could you please consider checking with Uday if it is OK to share specifics? I understand that Uday is together with you on the slide credits, and he is the founder of Sentry as well. The specifics will make it truly awesome for our audience.

            Warm Regards

            Vikas

            • Keshava Murali Elliadka
              By Keshava Murali Elliadka  ~  6 months ago
              reply Reply

              Hello Vikas,

                                   I discussed with Uday regarding this.We can share the  high level overview of the algorithm and the details of how the inference works on real-time. Hope this would give enough details for the audience to try out.

              However, we will not be able to share the particular architecture, the modifications and the coefficients we use.

              Please let me know if this is OK. Thanks in advance !

              Best,

              Keshava