The person behind this crowdsourced photo collection platform is asking the wrong question, and that is exactly why it is the right one. They are building a crowdsourced photo collection platform that auto-labels images with YOLO and CLIP, enriches each one with over 40 metadata fields like weather, GPS, and OCR, and they are smart enough to pause before choosing a direction. Most founders would have already picked a category, built the pipeline, and started pitching. Instead, they are asking the public what data feels missing. That is not indecision. That is the beginning of a product that might actually get used.
Here is what this means for you, whether you are a developer, a researcher, or someone who just wants better tools. The gap they are describing is real. European street scenes are not covered by any major dataset, and that is a genuine problem if you are building anything that needs to understand signage, storefronts, or pedestrian flow outside the US. Supermarket shelves with OCR-extracted prices sound niche until you realize how much consumer behavior research, price comparison apps, and supply chain analysis would benefit from having that data available on demand. Analog utility meters, restaurant menus, EV charging stations by type: every one of these is a piece of the physical world that remains invisible to the models we rely on daily. The fact that these are the ideas being floated tells you how much low-hanging fruit is still sitting there, unphotographed and unlabeled.
What makes this approach compelling is not the technology, though the tech is solid. It is the humility baked into the process. The platform is not pretending to be a magic box that will solve every data problem at once. It is asking contributors to point their smartphones at specific targets, then using automation to make the collection effort scalable. That is a practical division of labor: humans decide what matters, machines handle the tedious work of tagging and enriching. For you, this means the resulting dataset could be shaped by actual demand rather than by what a company assumes you want. You get a say in what gets captured first, and that is a rare thing in an industry where data collection is usually decided behind closed doors.
The concrete next step for the builder is to stop treating this as a polling exercise and start treating it as a discovery mechanism. Ask the community what they would use, yes, but also ask them what they would contribute. The people who answer the first question are not necessarily the ones who will go out and photograph a utility meter at 7 a.m. But the people who answer the second question are your real foundation. Build the feedback loop around contribution intent, not just wish lists. If the platform can show a small, useful dataset within a month, even if it is just 500 images of Swiss street scenes with weather and OCR data attached, that will tell you more than any thread of upvoted comments. The data you need is missing, but the path to it is not. It is in the answers to this post, and the next step is to start collecting before you have perfect clarity.