Question 1

What is the difference between this and standard 'Computer Vision'?

Accepted Answer

Standard Computer Vision focuses mostly on the algorithm (training a neural network to recognize a defective part). Custom Vision Solutions covers the entire deployment ecosystem: the physical camera setup, the edge computing hardware, connecting the AI to factory machinery, and building the dashboards.

Question 2

Can the AI interact with physical machinery?

Accepted Answer

Yes. In manufacturing settings, we can connect the edge AI device to Programmable Logic Controllers (PLCs). If the camera detects a defective product on the line, the AI can send an immediate digital signal to the PLC to halt the conveyor belt or activate a robotic arm to remove the item.

Question 3

Do you provide the cameras and hardware?

Accepted Answer

We are a software engineering firm, but we provide detailed hardware specifications (camera resolution, lens type, lighting requirements, and edge compute nodes) and partner with hardware vendors or your existing IT team for procurement and physical installation.

Question 4

How do you handle model drift (when the AI gets worse over time)?

Accepted Answer

Environments change (lighting degrades, new products are introduced). We build MLOps pipelines that continually sample the video feed, flag edge cases where the AI is uncertain, send them to a human for labeling, and automatically retrain and deploy an updated model.

Question 5

Is an internet connection required for this to work?

Accepted Answer

For the actual inference (detecting objects), no. By deploying models on edge devices, the system runs locally and in real-time even if the internet goes down. An internet connection is only required periodically to push model updates or upload aggregated metadata to the cloud.

Question 6

What happens if we have 50 different retail locations?

Accepted Answer

We implement fleet management tools. This allows us to monitor the health of all 50 edge devices from a central dashboard. When we improve the AI model, we can push the update Over-The-Air (OTA) to all 50 locations simultaneously, just like a software update on your phone.

Question 7

How do you ensure GDPR/CCPA compliance with video?

Accepted Answer

By processing the video locally on the edge device, the raw video feed never touches the cloud or a database. The AI extracts only the metadata (e.g., '1 person entered at 9:00 AM') and immediately discards the video frames, ensuring complete privacy compliance.

Question 8

Can we use our existing CCTV cameras?

Accepted Answer

Often, yes. We can capture the RTSP stream from your existing IP cameras and route it into our edge computing devices for AI processing, provided the camera angle and resolution are sufficient for the specific task.

How can we help?

Custom Vision Solutions
FAQ

Helpful
information

Getting Started

Timeline

Pricing

Support

Still have questions?

Custom Vision SolutionsFAQ

Helpfulinformation