OnDeck AI
Analyze any footage, without training a model
Hey everyone 👋
We’re Alexander and Sepand, cofounders of OnDeck AI.
TL;DR
OnDeck lets you analyze any video, without training a model. Instantly find any event, behaviour, or object without needing any training data.
Ask:
Do you know anyone who needs to analyze tons of video or imagery? Please introduce us! alexander@ondeck-ai.com
[https://youtu.be/MafiM94IkV8\]
The Pain:
Creating vision models can take months of engineering time: collecting training data, training, then deployment. Worse yet:
- it’s often impossible to get enough data for a specific task, and
- even the best cv models struggle to generalize across diverse camera setups, workflows and environments.
Our Magic:
To overcome these blockers, we bet early on the power of VLMs and built a vision engine that can generalize across any task and doesn’t need any training data. We published a NeurIPS workshop paper showing our new methods with VLMs beat traditional CV even at niche tasks.
Why this is good:
- We never train on customer data, unlocking sensitive use cases.
- Deploy instantly and skip months of R&D.
- Identify things that have never been seen before.
- Unlike traditional cv, OnDeck can understand qualitative behaviour and sequences of events.
This summer we’ve analyzed thousands of hours of footage across:
- Fleets of Autonomous Surface Vessels
- Robotics Research
- Security cameras
- Behaviour analysis for port monitoring
- Off-shore oil & gas monitoring
The Team:
Sepand (CTO) has a Master in ML from Cambridge and worked at the European Space Agency, Amazon, and swarm robotics research labs. We both studied cs in university together and we’ve been best friends since grade 6. We originally started OnDeck to automate fisheries monitoring (counting fish in video 🐠). We generated six-figure revenue before pivoting in January.
Ask: Do you know anyone with loads of video or imagery to analyze? Please introduce us!
Ping us anytime: alexander@ondeck-ai.com / ondeckai.com / Linkedin / Twitter
