KlinterAI - Integrating Observability in the Construction Industry

[#1]

Team Name	KlinterAI
Track	IoT
Use Case	Integrating Observability in the Construction Industry
Technology	Artificial Intelligence, Construction Software, Computer Vision Depth Camera - Vision as a Sensor, Observability using eBPF, Computer Programming Languages, BIM and ERP
Region	Worldwide
Team lead	Aswin Vijayakumar
Team members	Aswin Vijayakumar @aswinkvj, Afeef K K @afeef , LuisRodrigo @LuisRodrigo , Marcus Liles @marcusliles
Social media handles	https://www.linkedin.com/in/aswinvk28/ https://www.linkedin.com/in/afeef-k-k-438479a7/ https://www.linkedin.com/in/luis-rodrigo-b-887710187/ https://www.linkedin.com/in/marcus-liles-324021183/
Is your team interested in participating in an interview?	Y
Optional: Expand on your technology expertise	Works with AI, ML, Supplier-oriented Database, Computer Networking, IoT and Programming, Computer Vision and Deep Learning, Optimization and Statistics. A Construction Dataset is created from various Construction Foundation Sites, timestamped Digitally to perform Detection & Recognition on Machinery.

Jury Video:

Pitch Video:

RoboFlow Dataset

 

[Atabarut]

Computer Vision-Based Video Interpretation Model for Automated Productivity Analysis of Construction...

[Atabarut]

[akogokennedy]

Thanks @Atabarut I can't allow a cross talk system as said in: "It focuses on designing a mechanism for furthering the crosstalk between the prior knowledge of construction operations and computer vision techniques.". I have a set goal, which will be to complete that activity recognition model, by collecting data from construction sites, visits-based. It is foundation-related. This will be a big list of algorithms to apply based on the paper shared. Some of the statistics do involve, and I am considering it. Site walkthroughs / Scheduled Walks, are proposed for quality inspection and some of our competitors are doing it. 

Digital Twin is something, which can be applied only to large buildings, because in foundation sites, you wouldn't see anything but machinery, soil, construction piles and people. 

I've got a dataset, collected from 1995 onwards, from Dubai to London to other sites in the United Kingdom, I believe there is more. I've got only 14k+ images including some machinery certificates which I clustered using a Machine Learning clustering algorithm to separate people, documents / certificates and construction sites / vehicles. 

I'm linking my colleagues to the team, as they are currently busy. Looking forward to this, from hackathon team, or a mentor who can be assigned to us. We are based in UK (2), Ghana, India. 

[aswinkvj]

@Atabarut I don't see the content you shared online, it is a @SAS based software, where is this deployed currently and what do you use with such a platform? Is this beneficial for you, or are you running in a loss?

• There is no Activity Recognition model published involving Construction Sites, as BIM takes over most of the things
• Your model theoretically looks good, but I see that the Dog Robot there is just an Idea from the Boston team and is not deployed Enterprise-Wide. 
• BIM is a also a methodology and does not involve anything near to Environment & Material Science. 

Do you have an argument, here?

Best, 

Aswin

[Atabarut]

It's built with a partner so due to IP can't share the rest.

[Atabarut]

0002018RecognizingDiverseConstructionActivitiesinSiteImagesviaRelevanceNetworksofConstruction-Relate...

great example isn't it?

[aswinkvj]

@Atabarut the above content you've shares is not visible to me. 

[Atabarut]

I see you need to have a credential to check that out

Here is a detailed alternative thesis;

https://s3-eu-west-1.amazonaws.com/pstorage-purdue-258596361474/22869947/dissertation_JC_final.pdf?X...

[aswinkvj]

@KevinKalish we've had our pen ultimate meeting before April 12th, end of Cloud Service. 

[aswinkvj]

The video upload is taking a bit of time. Please bear with the upload.