Job Title

Job Description

About Signify​

Through bold discovery and cutting-edge innovation, we lead an industry that is vital for the future of our planet: lighting. Through our leadership in connected lighting and the Internet of Things, we're breaking new ground in data analytics, AI, and smart solutions for homes, offices, cities, and beyond.​

At Signify, you can shape tomorrow by building on our incredible 125+ year legacy while working toward even bolder sustainability goals. Our culture of continuous learning, creativity, and commitment to diversity and inclusion empowers you to grow your skills and career.​

Join us, and together, we’ll transform our industry, making a lasting difference for brighter lives and a better world. You light the way. ​

More about the role​

How we learn the people count sensor today

• Trial & error, experience based estimating where a line crossing and detection box should be defined
• This is an art, more than rule based
• Needs to be done for every sensor again, time consuming, not scalable

Scope

• Train a Recurrent Neural Network on labeled datasets, and the sensor will learn when people walk in or out, in all kinds of different situations.

Benefit

• No configuration required, the sensor is "smart" enough.

Why we believe it works

• The situation is always more or less the same: the sensor looking at a door or an entrance.
• The information that a RNN needs as input is available in the cloud,  The RNN inference can be cloud based.
• The latest RNN's are very powerful.  (e.g. bi-directional Long-Short Term Memory LSTM)

What is needed

• Large annotated dataset of people count sensor output sequences of people walking in and out of doors/entrances.

Approach for Proof of Concept

• Collect datasets of people walking in and out.  Annotate each set by a human.
• Mount a people count sensor above a door.
• Mount an occupancy sensor above the door, log the occupancy data
• Videotape the door 24/7
• Human checks the video tape every time the occupancy sensor was triggered and stores an annotated sensor data piece.
• Machine Learning engineer trains a RNN with the annotated dataset.
• Testing of the RNN inference happens with a separate dataset that is unseen by the RNN
• Test real-time inference
• If successful, implement inference in VBI

More about you

Background in machine learning.

Ideal for students looking for a graduation assignment in the direction of AI Technology Architect.

Everything we’ll do for you

List out benefits and anything else on offer for this role.

You can grow a lasting career here. We’ll encourage you, support you, and challenge you. We’ll help you learn and progress in a way that’s right for you, with coaching and mentoring along the way. We’ll listen to you too, because we see and value every one of our 30,000+ people.

We believe that a diverse and inclusive workplace fosters creativity, innovation, and a full spectrum of bright ideas. With a global workforce representing 99 nationalities, we are dedicated to creating an inclusive environment where every voice is heard and valued, helping us all achieve more together.