Image processing is rapidly growing, allowing unprecedented amounts of data analysis in many fields. We, the Digest Quest team, appreciated the challenge of jumping into this new field and seeing what we could come out with.

Of course, all of us carry around our own personal image processing machines, our smartphones. Why not design an app that can use this untapped processing potential to do something all of us try so far but fail to do: lose weight.

What it does

Digest Quest allows users to seamlessly pull out their phone, snap a picture of their food's nutritional label, and immediately see the fruits of their labor: a breakdown calorie by calorie of their diet, letting them leverage the knowledge of what they've consumed to improve their dieting habits.

How we built it

Digest Quest is a heuristic image processing (specifically, optical character recognition) web app. We utilize Tesseract's open-source JavaScript OCR API to turn a camera's picture into a smattering of characters vaguely related to what the nutritional label actually says. This is where our custom heuristic algorithm comes into play. Because OCR is not perfect, we apply representativeness heuristics to find the closest matches between the nutritional label's expected values and the values produced by Tesseract.

These values pulled from the label then go to populate a back-end database, and is visualized artistically to help users of all demographics understand their diets at a glance.

Challenges we ran into

The past 36 hours has been filled with roadblocks (educational, nonetheless). At first, we struggled with the DNS connecting to our host server provider. Once that was settled, we started down the long road of attempting to have JQuery and Node.js connect to our Tesseract API to allow for image processing, all of a web platform. While this was happening, we spent many hours reviewing our UX/UI design, as we found it essential to ensure an elegant user experience.

Accomplishments that we're proud of

DigestQuest leverages many exciting features. The first of these is our heuristic data parser, taking the garbled mess of an OCR API call and formatting it into a usable data base. We also proudly integrated many computer science concepts into one package, leading to some beautiful multi-language interoperability. Lastly, with our UX/UI emphasis and analysis, we were able to create a unique, one-of-a-kind approach to data visualization.

What we learned

We were, for the most part, new to JavaScript, setting up a DNS, and image processing APIs. We are proud to say to that we were able to tackle these new challenges (with help, of course), within a short period of time. We went into this hackathon with a desire to expand our skill-sets, and we definitely succeeded in this regard.

What's next for DigestQuest

We look forward to taking DigestQuest forward by optimizing our OCR analysis to enable recording of even more information on nutritional labels, allowing users to ensure they are consuming proper amounts of vitamins, iron, and other nutrients.

Built With

Share this project: