Using a Crowd Counting AI Model For Your Business Using a Crowd Counting AI Model For Your Business
Background Deep Learning is quickly bringing computer vision to a spot where businesses can start integrating powerful models in everyday use.... Using a Crowd Counting AI Model For Your Business


Deep Learning is quickly bringing computer vision to a spot where businesses can start integrating powerful models in everyday use. Some of the challenges that businesses had with adopting AI in the past — who will build and maintain these models, how do I deploy them, are they accurate enough are now addressed in a new platform from Unleashlive that aims to connect model developers to end users.

I am launching a crowd counting model on this AI platform. This platform will allow businesses to run their clips or images or live stream video through my model and get instant results without any headaches related to setting up an environment or GPU. You can now have a 1 hour free trial of this model at Unleash Cloud. First-time users please set up a new account.

A person detection/counting model is very versatile with applications in many areas. Here are a few I can think of and a small demo of the performance of my app using some videos from youtube.

Interesting Applications of People Counter

  1. Retail

Understanding foot traffic is very important in retail to organize merchandise in aisles, optimize store layout, understand peak times and potentially even protect against theft. You can now put a camera in your store and connect it to Unleashlive AI platform and get real-time data that can be ingested and analyzed. See the model in action below

People Counter — Retail

2. Infrastructure Planning

A lot of businesses or government agencies could use people counter to understand various things like how crowded are public places at a given time or how many people are using a particular street crossing every day etc.

Traffic Count

3. Safety

Another broad area of application can be safety. If you are searching for someone who is lost or if a business would like to ensure an area is human free before starting a big machine. In the image below my model could track a person who is trekking in the outback.

Safety application of people counter

More about the AI Model

The crowd count model used here is a Faster RCNN Inception model trained on a custom dataset of people at various scales. It works faster and has higher accuracy than using a model trained on the COCO dataset. It works well on images of the crowd taking from close by and from very far up, as well as under different background and lighting conditions.

I want to use this model for my business

If you are intrigued by this article and would like to try the people counting model, then please sign in and create an account at https://cloud.unleashlive.com/

You will be able to active this model and test it out! And the best part is Unleash is offering a free 1 hour trial to all new users 🙂

I’m a developer and I want to add a model

This is easy too. Simply create an account on Unleash live and enable the developer flag (‘start as a developer’) in your profile in the top accordion. This sets up a private Sandbox for your testing. Having experience with tensorflow 1.6 and object detection is a plus.

Once you have enabled your account, ‘Developer’ will appear on the top navbar. Simply click and choose ‘Sandbox’ and follow the instructions provided. It is a very easy process.

Find out More

To learn more about the Unleashlive platform and how the business model works, please check out the FAQs below:


If you have questions or need help, then feel free to contact us at getstarted@unleashlive.com. If you have ideas for other AI apps, please add those in comments below.


Priya Dwivedi

Priya Dwivedi

Priya Dwivedi has 10+ years experience as a data scientist. She now runs her own data analytics consultancy that builds deep learning models for Computer Vision and NLP problems. She has helped many startups deploy innovative AI based solutions. For more info please see the link — http://deeplearninganalytics.org/. If you are interested in collaborating with her then please contact her at priya.toronto3@gmail.com.