Urgent.News

One page, thousands of outlets. See who else covered it.

Editions

AI

Stop Guessing Calories: Build a Multimodal Food Estimation Pipeline with GPT-4o & SAM

We’ve all been there: staring at a delicious plate of pasta, trying to figure out if it's 400 or 800 calories. Manual tracking is a chore, and standard apps often fail at portion estimation. But what if we could combine Computer Vision , Multimodal LLMs , and Vector Databases to build an automated nutritionist? In this tutorial, we are building a state-of-the-art Multimodal Food Estimation…

In a world where estimating the calorie content of a food item can often feel like a guessing game, a new automated system has emerged to provide precise nutritional analysis. By combining the power of Computer Vision, Multimodal Large Language Models (LLMs), and Vector Databases, researchers have developed a state-of-the-art Multimodal Food Estimation Pipeline.

This innovative approach uses the Segment Anything Model (SAM) to precisely isolate food items within an image and directs GPT-4o Vision to perform contextual analysis, bridging the gap between visual perception and nutritional calculations. The pipeline employs a structured Identify-Analyze-Match flow, ensuring that the LLM focuses on relevant pixels and minimizing the risk of hallucinations.

After analyzing the image, the system retrieves verified nutritional data using a Postgres database with the pgvector extension. This combination of advanced technologies aims to revolutionize the way we track and understand the nutritional content of our meals, offering a sleek and efficient solution for anyone interested in AI-driven wellness, FastAPI development, or Multimodal Retrieval Augmented Generation (RAG).

Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

Read the original at dev.to →

More in AI

One terminal, two trust levels — running Claude Code against a real subscription and a cheap proxy

Part of an ongoing series on model routing and trust tiering for agentic coding tools. This one's the boring, working half — no bug hunt, just a setup that's been running clean across two machines.

  • Claude Code runs on two trust levels using a real subscription and a cheap proxy
  • Self-hosted proxy translates Anthropic-format requests to DeepSeek V4
  • Proxy config separates planning/review tasks from high-volume, low-stakes work

Why AI Agent Runtimes Need a 'Constitution': Lessons from Ironclaw and the Rise of Policy-First Autonomous Systems

Originally published on tamiz.pro . Introduction Autonomous AI agents are transitioning from research prototypes to production-critical systems.

  • Autonomous AI agents perform critical tasks requiring safety measures.
  • Constitution for AI agent runtimes addresses alignment and reliability issues.
  • Ironclaw runtime case study demonstrates challenges and policy-first solutions.

State agencies to ramp up AI usage

Five state agencies plan to adopt agentic artificial intelligence (AI) to support their complex tasks, from detecting foreign nominees buying land in Thailand to identifying fake parentage in civil…

  • Five Thai state agencies to boost AI usage
  • Agencies to use agentic AI for complex tasks
  • Aim to improve efficiency and accuracy

🐾 PawSafe: An AI-Powered Food Safety Checker for Dogs

This is a submission for Weekend Challenge: Dog Days Edition What I Built PawSafe is an AI-powered web application that helps dog owners answer a simple but important question: "Can my dog eat this?"…

  • PawSafe AI app helps dog owners check food safety.
  • Users input food name, upload image, or both for analysis.
  • Google Gemini API categorizes safety into four levels.

More from Monday 17 August →