April 27, 2021
Where to start with AI when you don’t have imagery or video
At our core, we want computer vision (CV) to be useful to your business or organization. What happens if you don’t have imagery and video but already see the value from automating certain manual, visual tasks? Most of our customers come to us early in the CV lifecycle: they have a problem and think AI can help solve it.
Frankly, this is the best time to engage: before data collection diverges from subject matter experts’ needs and expectations.
A large portion of our in-bounds fall into these two groups:1. Users who don’t have imagery and want to learn more about how to start a computer vision project.
2. Users who already have some imagery, but it's limited or focused on one particular domain.
This post is for people in group 1.
(Don’t worry group 2: we’ll have a post for you soon!)
You may be saying “I don’t have imagery or video but want to use computer vision” or “I see my competitors using computer vision to automate workflows, and I want to, too.”
There is nothing stopping you from accomplishing those things! You may not have imagery yet, but just think about all the other departments in your business that do. Many enterprises collect some type of imagery or video as part of their daily routine work. Below are a few examples based on our experience—ask yourself if your organization collects any of these.
- Cell phone cameras
- Open-source satellite, aerial, or drone footage
- Security cameras / CCTV
- Gel imaging or microscopy
- Field-deployed laptops
- User or customer-uploaded imagery
Your business probably does have something similar to these right? More often than not, imagery and video are collected in different parts of the business that may not touch the customer. Most large facilities and companies collect security camera footage. But certain video and imagery collections may not be persistent and may be deleted or put in cold storage after a certain amount of time.
Use what you have.
Speak with the other departments, regions, or sectors of your business. Are they collecting or sourcing imagery that could be repurposed for your use-case? Look internally before spending the time and money to create new camera systems. This may require gumption and creativity, but you may be surprised to find just how much imagery you’re already collecting for other purposes!
If your business truly does not have usable imagery and videos, think about the types of visual sensors that you have that might be available to you. Do your colleagues carry company-sponsored smartphones? Do you have security cameras or smart doorbells? How about dashcams on company vehicles? CrowdAI is happy to explore with you which hardware you could use to start collecting imagery. (Click here to set up a talk with one of our AI experts.)
Take the first step to find your existing imagery and you’ll be surprised to find that you have an abundant amount that we can help you turn into a model.
Pablo is a co-founder of CrowdAI and leads Data Operations.
March 24, 2020
Tornadoes are unpredictable. Unlike other storms, where we m....
August 22, 2018
Hello, world! We are CrowdAI, a deep learning company proudl....
May 06, 2021
Previously, we discussed the two important reasons as to why....