M12 lenses for IoT and smart retail cameras

Compact S-mount optics for intelligent checkout, shelf and inventory vision, people counting, and queue analytics.

Smart-retail cameras sit on ceilings, gondola rails, and self-checkout kiosks. The lens has to stay small, hold focus through the temperature swing of a cold aisle, and cover the scene the analytics model actually needs. People counting wants an ultra-wide fisheye that takes in a whole zone from one ceiling mount. Reading shelf labels and checkout barcodes wants a low-distortion 3mm to 6mm that keeps text straight out to the frame edge. We stock 18 M12 lenses for this work, from a 230° fisheye to a 6mm at F/1.7, and single samples ship same day from San Diego on orders placed by 12 PM PST.

18 Retail-vision M12 lenses
1.8–6mm Rectilinear EFL range
190°–230° Fisheye FOV options

Lenses grouped by retail vision task

Start from the job the camera does. Each pick carries the field of view and focal length from its own datasheet.

Wide-area coverage and people counting

Ceiling and dome cameras that count people, measure dwell, and watch a whole sales floor run on fisheye optics. One wide field covers a zone from a single mount, and the heavy barrel distortion is not a problem, because the counting model works from calibrated geometry rather than a picture a person looks at.

Self-checkout, shelf, and OCR

Barcode reading, planogram checks, and self-checkout vision need straight lines. A low-distortion or no-distortion rectilinear lens keeps characters and shelf edges true across the frame, so the OCR and detection stages do not fight lens warp. Focal length sets how much counter or shelf each camera sees.

Compact and specialty

Small footprints for kiosks and embedded boards, plus IR correction for lighting that changes through the day and a large-format option that fills bigger sensors.

Match a lens to the retail task

Six common smart-retail jobs and the lens that fits each one.

Retail vision task おすすめのレンズ EFL 視野 この選手を選んだ理由
Overhead people counting, wide floor zone 195°@4.7mm Fisheye M12 4.7mm 195° design field One ceiling camera covers a whole zone. The counting model uses calibrated geometry, so barrel distortion does not hurt accuracy.
Single-camera store surround 230°@4.5mm Fisheye M12 4.5mm 230° design field Widest field in the set. One sensor takes in a full room where mounting points are scarce.
Self-checkout barcode and item OCR Low Distortion 3.0mm M12 3.0mm 91° HFOV on AR0234 Low distortion keeps barcodes and characters straight. The field is wide enough to see the full bagging area up close.
Shelf and planogram reading, cold aisle No Distortion 3.2mm IP67 3.2mm 83° HFOV on AR0234 The no-distortion field keeps shelf-label lines true to the edge. IP67 sealing shrugs off condensation and washdown.
Queue and entrance counting at standoff Low Distortion 6mm M12 6mm 53° HFOV on AR0234 A narrower field puts more pixels on people several meters away. F/1.7 and IP67 suit dim vestibules and outdoor doors.
Day-to-night entrance under IR light 190°@4.2mm IR Corrected Fisheye M12 4.2mm 190° design field IR correction holds focus when the scene switches from daylight to near-infrared illumination at night. The wide field covers the doorway.

Rectilinear field of view is the horizontal FOV computed for the AR0234 (1/2.6-inch, 1920 x 1200) by the Commonlands field of view backend. Fisheye figures are the lens design field from the datasheet, and the imaged field narrows on sensors smaller than the image circle. Check any lens against your own sensor with the field of view calculator and the sensor size chart.

How to choose an M12 lens for smart-retail cameras

Retail vision splits into two optical problems. One is coverage: see the whole floor or aisle from as few cameras as possible. The other is legibility: put enough clean pixels on a barcode, price label, or face for the model to read it. Fisheye and rectilinear lenses sit at opposite ends of that split, and most store builds use both.

Field of view versus reading detail

Coverage tasks reward wide angles. A 190° to 230° fisheye on a ceiling covers a zone that would take three or four narrow cameras, which is why people-counting and heatmap systems lean on them. Reading tasks reward the opposite. Every degree of field you add spreads the same sensor pixels over more scene, so a barcode or shelf label lands on fewer pixels. Pick the longest focal length that still frames the counter or bay, not the widest one that fits.

Distortion is a legibility spec, not a looks spec

Barrel distortion (negative TV distortion) bows straight lines outward and stretches shapes near the frame edge. For a person-counting fisheye that is fine, because the analytics run on calibrated geometry. For OCR it is a problem: characters and barcode bars near the edge shear and the decode rate drops. Low-distortion and no-distortion designs hold lines straight across the field. Software can undistort a wide lens, but only as well as its calibration, and it adds a step to every frame.

Sensor coverage and image circle

A lens only fills a circle of light, and that image circle has to reach the corners of your sensor or they go dark. The retail lenses here run 8.1mm to 9.4mm image circles, which fully cover a 1/2.6-inch machine-vision sensor such as the AR0234 (6.8mm diagonal). For a sensor larger than that, the large-format 5.5mm (CIL056) is built to fill a bigger frame. When in doubt, run the pairing through the calculator, which flags coverage before you buy.

Low light, IR, and 24/7 operation

Entrances and stockrooms run around the clock, so aperture and IR behavior matter. A fast lens such as the 6mm at F/1.7 (CIL059) gathers more light per exposure than a slower one. Cameras that switch to near-infrared illuminators at night want an IR-corrected lens: it holds focus between visible and near-infrared, where an ordinary lens shifts focus and softens. The 190° IR-corrected fisheye (CIL215) and the IR-corrected 4.4mm (CIL046) are built for that switch.

Sealing and environment

Cold aisles, vestibules, and outdoor-facing entrances see condensation, dust, and washdown. IP67-rated lenses such as the 3.2mm (CIL034) and 6mm (CIL059) handle that without a separate housing, which saves space on a compact camera and removes a fogging surface in front of the optics.

Mount and integration

M12 is a 12mm x 0.5mm thread, also called S-mount. It fits M12 camera modules from ArduCam, e-con Systems, and Leopard Imaging, and the boards that carry AR0234, IMX290, and similar sensors on NVIDIA Jetson and Raspberry Pi. Match the flange and back focal length to your module, confirm the image circle covers the sensor, and set focus against the actual scene distance before you lock the thread.

どのレンズにしようか迷っていますか?

Send us your sensor part number and mounting distance. We will point you to the field of view and focal length that fit.

エンジニアリング部門へのお問い合わせ

Common questions from retail vision teams

What engineers ask most when they spec lenses for smart-store cameras.

Which lens should I use for ceiling people counting?

Use a fisheye in the 190° to 230° range so one ceiling camera covers a whole zone. The 195° at 4.7mm and the 230° at 4.5mm are the usual picks. Barrel distortion is fine here, because the counting model runs on calibrated geometry rather than a straightened image.

What focal length reads shelf labels and barcodes?

It depends on aisle width and how far the camera sits from the shelf, but a low-distortion 3mm to 6mm covers most cases. The 3.0mm (CIL030) gives a 91° horizontal field on the AR0234 for close counters, and the 6mm (CIL059) narrows to 53° for more pixels at standoff. Confirm the framing against your sensor with the field of view calculator.

Why does lens distortion matter for OCR?

Barrel distortion bows straight lines and stretches shapes near the frame edge. Barcode bars and label characters that land in that region shear, and the decode or read rate falls. A low-distortion or no-distortion lens keeps lines straight across the field. Software can correct a distorted image, but the result is only as good as the calibration, and it adds work to every frame.

Do these fit Jetson, Raspberry Pi, and ArduCam retail boards?

Yes. M12 is a 12mm x 0.5mm thread that fits M12 modules from ArduCam, e-con Systems, and Leopard Imaging, including boards that carry AR0234, IMX290, and similar sensors on Jetson and Raspberry Pi. Check that the lens image circle covers your sensor diagonal. Our sensor size chart lists the diagonals.

Can I run these at entrances and in cold aisles?

Yes. The IP67-rated 3.2mm (CIL034) and 6mm (CIL059) handle condensation, dust, and washdown without a separate housing. For entrances that switch to infrared light at night, an IR-corrected lens such as the 190° fisheye (CIL215) holds focus between visible and near-infrared.

How fast do samples ship?

Single-lens samples ship the same day from San Diego on orders placed by 12 PM PST, with no minimum order. For a rollout across a fleet of cameras, ask us to pull the lenses from one production lot so unit-to-unit variation stays low across the deployment.

Building a smart-retail camera?

We work with retail analytics and self-checkout teams from first prototype to rollout. Single-lens samples ship same day. No minimum order.