https://www.intsoftsop.com/en/ Intsoft (Shenzhen) Technology Co., Ltd. builds Visual AI SOP Systems that use industrial machine vision to recognise workstation actions in real time, delivering SOP compliance inspection, error prevention, missing-process alarms and production traceability. en_US Wed, 21 Oct 2026 03:59:52 +0800 https://www.intsoftsop.com/en/blog/visual-sop-vendor-evaluation.html https://www.intsoftsop.com/en/blog/visual-sop-vendor-evaluation.html Tue, 06 Oct 2026 17:55:00 +0800 The differences between vision suppliers are hard to see in promotional material or in a quotation. Every supplier has a list of case studies, every supplier can use the terminology, and the differences in price usually reflect the specification rather than the capability. The place where capability… https://www.intsoftsop.com/en/blog/lighting-aging-and-degradation.html https://www.intsoftsop.com/en/blog/lighting-aging-and-degradation.html Tue, 06 Oct 2026 17:50:00 +0800 Lighting receives a great deal of attention when the scheme is designed and very little afterwards, where it is treated as a consumable. But its output changes, and the way it changes is more complicated than most people assume: not a slow linear dimming, but a fast initial fall followed by a gentle… https://www.intsoftsop.com/en/blog/vision-project-risk-register.html https://www.intsoftsop.com/en/blog/vision-project-risk-register.html Tue, 06 Oct 2026 17:45:00 +0800 Vision projects fail in repetitive ways. Setting them out as a register and working through it at proposal and scheme review is far more useful than reconstructing what went wrong afterwards. Nothing in this list is unusual: each item is something most projects encounter to some degree. The differen… https://www.intsoftsop.com/en/blog/cross-department-roles.html https://www.intsoftsop.com/en/blog/cross-department-roles.html Tue, 06 Oct 2026 17:40:00 +0800 A vision project passes through process engineering, quality, maintenance, production, IT and purchasing, plus an outside supplier. The technical problems generally have known answers. What stalls the project is a gap in accountability — something that everybody assumed somebody else was handling, u… https://www.intsoftsop.com/en/blog/sample-library-management.html https://www.intsoftsop.com/en/blog/sample-library-management.html Tue, 06 Oct 2026 17:35:00 +0800 A sample library is the foundation under a vision system. It proves at acceptance that the system can do the job, it monitors day to day whether the system still does it, and it settles disputes about whether the system or the criteria misjudged a part. It is also the asset that disappears most easi… https://www.intsoftsop.com/en/blog/audit-evidence-preparation.html https://www.intsoftsop.com/en/blog/audit-evidence-preparation.html Tue, 06 Oct 2026 17:30:00 +0800 When a customer or a certification body reviews a vision station, the questions are predictably shaped: who decided what acceptable means, does the system actually do that, does anybody confirm it still does, and what has been changed since. Nearly all of it can be answered from records that exist —… https://www.intsoftsop.com/en/blog/criteria-alignment-training.html https://www.intsoftsop.com/en/blog/criteria-alignment-training.html Tue, 06 Oct 2026 17:25:00 +0800 Anyone can write down what "acceptable" means. Getting a group of people to apply it the same way, on the same part, on different days, is a different problem. The usual approach — read the document aloud, sign the sheet — produces almost nothing, because the difficulty has never been the wording. I… https://www.intsoftsop.com/en/blog/daily-check-and-maintenance.html https://www.intsoftsop.com/en/blog/daily-check-and-maintenance.html Tue, 06 Oct 2026 17:20:00 +0800 Vision systems fail in two ways. Some failures are abrupt — a camera drops off the network, software errors out, the program exits — and the line notices immediately without anybody checking. The rest are gradual: a light dimming, a window clouding over, a locating face wearing, a calibration drifti… https://www.intsoftsop.com/en/blog/inference-deployment-and-optimisation.html https://www.intsoftsop.com/en/blog/inference-deployment-and-optimisation.html Tue, 06 Oct 2026 17:15:00 +0800 Latency on a vision station is not the time the model takes. It is the whole path from the trigger arriving to the result leaving: image transfer, preprocessing, inference, post-processing, publication. Breaking it down is the first step and the one most often skipped, because the intuition that the… https://www.intsoftsop.com/en/blog/vision-guided-positioning.html https://www.intsoftsop.com/en/blog/vision-guided-positioning.html Tue, 06 Oct 2026 17:10:00 +0800 To a customer, "use a camera to find where the part is" and "use a camera to check whether the part is good" sound like the same request. They are two different disciplines. One produces a continuous coordinate delivered into a control loop; the other produces a discrete verdict that decides whether… https://www.intsoftsop.com/en/blog/multi-camera-coordinate-alignment.html https://www.intsoftsop.com/en/blog/multi-camera-coordinate-alignment.html Tue, 06 Oct 2026 17:05:00 +0800 A single camera station needs one calibration. The moment a second camera is added, a second problem appears: each camera is accurate in its own frame, and the two frames differ by a transform nobody has established. That transform has to be built from something both cameras can see, or from a mecha… https://www.intsoftsop.com/en/blog/hand-eye-calibration.html https://www.intsoftsop.com/en/blog/hand-eye-calibration.html Tue, 06 Oct 2026 17:00:00 +0800 The camera reports pixels; the robot needs millimetres and an angle. The bridge between them is a calibration built from a set of positions where both frames were recorded, and it is the single most important and least visible step in any guidance project. Wrong calibration raises no alarm. The robo… https://www.intsoftsop.com/en/blog/strobe-and-continuous-lighting.html https://www.intsoftsop.com/en/blog/strobe-and-continuous-lighting.html Tue, 06 Oct 2026 16:55:00 +0800 The choice between continuous and strobe lighting is a choice about where the light comes from at the moment of exposure. Continuous lighting runs at a modest, sustainable level all the time. A strobe drives the same emitter hard for a very short interval, delivering far more light during the exposu… https://www.intsoftsop.com/en/blog/camera-protection-and-cleaning.html https://www.intsoftsop.com/en/blog/camera-protection-and-cleaning.html Tue, 06 Oct 2026 16:50:00 +0800 A camera on a production line is exposed to whatever the process throws at it: dust from casting or grinding, oil mist from machining, coolant spray, weld spatter, washdown water, cleaning chemicals. Optical surfaces are the most vulnerable part of the station and the least tolerant of contamination… https://www.intsoftsop.com/en/blog/line-scan-camera-basics.html https://www.intsoftsop.com/en/blog/line-scan-camera-basics.html Tue, 06 Oct 2026 16:45:00 +0800 A line scan camera has a sensor that is one row of pixels, or a few rows, and it builds an image by taking a picture of a thin slice of the product over and over as the product moves past it. The resulting image is a composite whose geometry depends on how fast the product was moving. This makes the… https://www.intsoftsop.com/en/blog/depth-of-field-and-aperture.html https://www.intsoftsop.com/en/blog/depth-of-field-and-aperture.html Tue, 06 Oct 2026 16:40:00 +0800 Depth of field is the range of distances over which the image stays acceptably sharp. It matters to a vision station because parts are never perfectly flat and never perfectly in place: a measurement taken on a surface that has drifted out of the sharp range is a measurement taken on a blurred edge,… https://www.intsoftsop.com/en/blog/inspection-record-retention.html https://www.intsoftsop.com/en/blog/inspection-record-retention.html Tue, 06 Oct 2026 16:30:00 +0800 A traceability claim is only worth what the record behind it is worth. Asked to prove that a particular part was inspected and passed, a site either produces the evidence quickly or discovers that the images were overwritten, the values were summarised, and the link between the part and the record w… https://www.intsoftsop.com/en/blog/inspection-data-to-spc.html https://www.intsoftsop.com/en/blog/inspection-data-to-spc.html Tue, 06 Oct 2026 16:00:00 +0800 A pass or fail verdict is the end of a decision and the beginning of a datum. The station measured something to reach that verdict — a distance, an angle, a grey-level difference, a position offset — and that measurement is far more useful to the process than the binary outcome, because it moves bef… https://www.intsoftsop.com/en/blog/defect-sample-scarcity.html https://www.intsoftsop.com/en/blog/defect-sample-scarcity.html Tue, 06 Oct 2026 15:30:00 +0800 A good process produces few defects, which is precisely the problem: the defect you most need to catch is the one with the fewest examples, and often none at all. Adding data is not simply a matter of waiting, because a rare failure may not occur again for months and may occur in a form nobody photo… https://www.intsoftsop.com/en/blog/deep-learning-vs-rule-based-vision.html https://www.intsoftsop.com/en/blog/deep-learning-vs-rule-based-vision.html Tue, 06 Oct 2026 15:00:00 +0800 The choice is usually presented as a matter of ambition, with deep learning as the modern option and classical image processing as the legacy one. That framing produces bad projects in both directions: simple gauging tasks wrapped in a model that now needs retraining, and appearance judgments attack…