Computer Vision Developer Needed – Fix Web-Based PD Measurement Accuracy

I’m looking for an experienced developer who can help us fix an existing web-based PD (pupillary distance) measurement tool and then continue development of the application. The app is currently built around Laravel, and the main issue we need to solve right now is the **accuracy and consistency of the PD measurement**. We have been testing the current version with real users and have found some concerning inconsistencies. For example, one tester has a known PD of **58.5 mm**, but the application is consistently measuring around **63 mm** on an iPhone 14 using Chrome. Another tester has noticed that the application requires the user's face to be relatively far away from the screen before it takes the measurement. His concern is that when the face occupies a smaller portion of the screen, there may be fewer pixels/landmarks available for the system to accurately calculate the distance. The current testing has also been done across different browsers and modes, including clearing cache, using different browsers, and incognito mode, so we're looking beyond simple browser issues. What I need from you I need someone who can **properly investigate the current measurement logic**, identify why the results are inaccurate/inconsistent, and determine the best technical approach to improve the accuracy. I'm not looking for someone to simply tweak a few numbers until the result looks right. You should be comfortable working with things such as: * Facial landmark / face detection technology * Computer vision * Camera-based measurements * Browser camera APIs * iPhone/Android camera behavior * Image scaling and calibration * Distance/depth estimation * JavaScript * Laravel/PHP If the current approach can be properly calibrated and improved, great. If you believe the current approach is fundamentally limiting the accuracy, I want you to be able to explain why and recommend a better approach or technology. There are also existing solutions/tools that we can use as reference, including Optogrid, Jeeliz, Banuba, Perfect Corp, Fittingbox, etc. You don't necessarily need to use any of these, but I want you to be familiar with evaluating existing computer-vision solutions rather than automatically building everything from scratch. Important I want the measurement tested against **known PD values** so we can objectively determine accuracy. For example: Known PD: 58.5 mm App result: 63 mm Error: +4.5 mm The goal is to get the measurement as accurate and repeatable as realistically possible across different devices and users. Once the measurement component is properly addressed, there is additional Laravel development that needs to be completed on the application. What I'm looking for I'm looking for someone who can actually **diagnose the problem**, not just follow instructions. Please tell me: * Your experience with camera-based measurement or computer vision * Similar projects you've worked on * What technologies you've used for facial landmarks/face detection * Whether you've worked with browser-based camera applications * How you would approach diagnosing the current PD accuracy issue If you have examples of similar work, please include them. This is an existing application, so you will be working with an existing codebase rather than starting completely from scratch. I'm looking for someone who can take ownership of the technical problem, communicate clearly, and recommend the right approach rather than simply doing whatever is easiest. Please don't apply if your experience is limited to basic Laravel/PHP development. The **PD measurement/computer vision portion is the most important part of this project.**