{
  "id": 1626458,
  "title": "Building OopsCalorie: When Your AI Thinks Dinuguan Is Champorado 😂",
  "url": "https://urgent.news/2026/08/18/building-oopscalorie-when-your-ai-thinks-dinuguan-is-champorado",
  "topic": "ai",
  "section": "AI",
  "published": "2026-08-18T03:29:45.000Z",
  "source": {
    "name": "Dev.to",
    "slug": "dev-to",
    "url": "https://dev.to/chimkennuggyszdev/building-oopscalorie-when-your-ai-thinks-dinuguan-is-champorado-4961"
  },
  "original_language": "en",
  "account": "I recently started working on a side project called OopsCalorie, an AI-powered calorie and meal tracking app. At first glance, the concept seemed straightforward: users log or take a photo of their food, the AI identifies the meal, estimates calories and macronutrients, and saves the entry. But when we began testing with Filipino dishes, our AI made some hilariously inaccurate connections.\n\nOne example was when the AI confidently identified dinuguan as champorado - both dark dishes served in bowls, but clearly distinct from one another. Another funny mix-up was the AI mistaking bagnet for Lumpiang Shanghai. These misidentifications may be amusing, but they reveal a deeper engineering challenge: accurately recognizing regional Filipino dishes requires more than just image recognition.\n\nEstimating calories is only the first step. The app must correctly identify the food item, which can be ambiguous based on the photo alone. A single image may show multiple dishes, hidden ingredients, visually similar foods, regional dishes underrepresented in training data, different cooking methods, unknown portion sizes, or ingredients concealed by other components. Even humans need context to make these determinations.\n\nTo address these issues, OopsCalorie is evolving to treat AI estimates as intelligent suggestions rather than absolute truths. Users can provide context through a brief description and the system will consider multiple food candidates instead of definitively naming one. The current tech stack includes React Native, NestJS, MongoDB, and LLMs/multimodal AI. The goal is to streamline food logging by allowing users to take a photo, add context if needed, and receive an estimate, with the option to correct the AI's guess when necessary.\n\nI invite developers, designers, nutrition/fitness enthusiasts, and anyone interested in experimenting with AI-powered consumer apps to join as beta testers. You can follow OopsCalorie on their website, become a beta tester, or contribute to development discussions on their blog.",
  "summary": "I’ve been building a side project called OopsCalorie , an AI-powered calorie and meal tracking app. The idea sounded simple enough: User logs or takes a photo of their food. AI identifies the meal. Estimate calories and macros. Save the entry. Done. Simple, right? Well... Then we started testing it with Filipino food. 😂 AI Meets Filipino Food 🇵🇭 One of the funniest parts of building…",
  "key_points": [],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}