How do you calculate machine vision image storage?

Multiply images per hour by average file size by the retention period to get raw volume, then multiply by the compression factor. With 300 images/hr at 5 MB over 720 hours and a 0.15x factor, raw volume is 1,500 (in the formula's scaled units) and compressed storage works out to 162,000 MB — about 158 GB.

What compression factor should I use for inspection images?

A 0.15x factor means compressed files are 15% of uncompressed size — typical for lightly textured monochrome parts saved as good-quality JPEG. Lossless PNG often lands near 0.4–0.6x, while aggressive JPEG on flat backgrounds can reach 0.08x. Validate on a sample of your actual frames rather than assuming.

How much storage does 300 images per hour need for a month?

At 5 MB per uncompressed image over a 720-hour retention window (about 30 days) with a 0.15x compression factor, this calculator returns 162,000 MB, roughly 158 GB of compressed data on disk.

Should I store every image or only rejects?

Storing every frame maximizes traceability and training data but multiplies cost. Many shops keep all rejects plus a rolling sample of good parts. If you save only rejects, replace images per hour with your reject rate times throughput before running this calculator.

Does JPEG compression hurt defect detection?

It can. Lossy compression introduces blocking artifacts that look like defects or hide hairline cracks. For archival and human review a 0.15x JPEG is usually fine; for retraining a model, keep a lossless copy of borderline images even though it raises storage.

Image storage vs network bandwidth — which limits me first?

Storage caps total retention; bandwidth caps capture rate. At 300 images/hr the sustained write is modest, but a burst inspection saving 50 frames in one second can saturate a 1 GbE link well before the disk fills.

Machine Vision & Industrial Inspection AI calculator

Vision Image Storage Calculator

Vision Image Storage estimates how much disk or NAS capacity a machine vision inspection line will consume over a retention window. Quality engineers and controls integrators use it when an inline camera saves a frame for every part — or every reject — and those images must be archived for traceability, false-reject review, or model retraining. Because a single 4K monochrome frame can be 5 MB and a line may shoot hundreds of images per hour, raw volumes balloon fast, and storage is where most vision projects silently overrun budget. This calculator converts capture rate, frame size, retention, and a compression ratio into a defensible MB figure before you size the array.

What this calculator does

Estimate the total image storage volume required for a machine vision inspection system based on image capture rate, file size, retention period, and compression ratio, so you can size storage infrastructure before deployment.
Use it when designing storage infrastructure for a machine vision system that retains inspection images for traceability, AI training data collection, or audit purposes.
It computes the compressed storage volume in MB that captured inspection images will occupy over the chosen retention period.

Formula used

Raw image volume = images per hour x file size x retention period
Compressed storage volume = raw image volume x compression factor

Inputs explained

Images captured per hour:
Average image file size (uncompressed):
Retention period:
Compression factor:

How to use the result

Use it when sizing local or networked storage for an inline vision system that archives frames for traceability, audit, or training-data collection.
It assumes a single average frame size and a constant compression ratio; mixed resolutions, color depth changes, or saving only on reject will shift the real figure materially.

Common questions

How do you calculate machine vision image storage? Multiply images per hour by average file size by the retention period to get raw volume, then multiply by the compression factor. With 300 images/hr at 5 MB over 720 hours and a 0.15x factor, raw volume is 1,500 (in the formula's scaled units) and compressed storage works out to 162,000 MB — about 158 GB.
What compression factor should I use for inspection images? A 0.15x factor means compressed files are 15% of uncompressed size — typical for lightly textured monochrome parts saved as good-quality JPEG. Lossless PNG often lands near 0.4–0.6x, while aggressive JPEG on flat backgrounds can reach 0.08x. Validate on a sample of your actual frames rather than assuming.
How much storage does 300 images per hour need for a month? At 5 MB per uncompressed image over a 720-hour retention window (about 30 days) with a 0.15x compression factor, this calculator returns 162,000 MB, roughly 158 GB of compressed data on disk.
Should I store every image or only rejects? Storing every frame maximizes traceability and training data but multiplies cost. Many shops keep all rejects plus a rolling sample of good parts. If you save only rejects, replace images per hour with your reject rate times throughput before running this calculator.
Does JPEG compression hurt defect detection? It can. Lossy compression introduces blocking artifacts that look like defects or hide hairline cracks. For archival and human review a 0.15x JPEG is usually fine; for retraining a model, keep a lossless copy of borderline images even though it raises storage.

Last reviewed 2026-05-12.