🐶 Can AI Understand Your Dog? Meet Dogesh — Your Dog, Decoded.

*This is a submission for Weekend Challenge: Dog Days Edition

What I Built

Dogesh is an AI-powered dog behavior interpreter that helps people understand what their dog may be communicating through visible body language.

Users can upload a photo of a dog, and Dogesh uses Google Gemini to analyze the image for visible clues such as facial expression, eyes, ears, mouth, posture, and overall body language.

Instead of returning a fixed or fake mood, Dogesh generates a dynamic analysis based on the uploaded image. It provides:

• 🐶 The most likely current emotional state
• 📊 AI confidence
• 🐾 Visible behavioral clues
• ⚠️ A warning when visible behavior may suggest fear, aggression, discomfort, or possible illness

Dogesh is designed as a fun but responsible way to help people look more closely at their dog's body language. It is visual interpretation, not a medical diagnosis.

Demo

https://youtu.be/Q5eiGpyXtBY?si=OKrFHSSMOSKLciAQ

You can also include a short screen recording showing:

1. Uploading a dog photo
2. Clicking "Analyze this dog"
3. Gemini analyzing the image
4. Dogesh displaying the detected state, confidence, visible clues, and warning

Code

https://github.com/AI-Explorer786/dogesh-your-dog-decoded

How I Built It

Dogesh was built with Next.js, TypeScript, and Google Gemini.

The application uses a simple image-analysis pipeline:

1. The user uploads a dog photo through the frontend.
2. The selected image is sent to a Next.js API route.
3. The API converts the image into a format that can be provided to Gemini.
4. Google Gemini analyzes the actual uploaded image.
5. Gemini evaluates visible features including facial expression, eyes, ears, mouth, posture, tail when visible, and overall body language.
6. The model returns structured JSON containing the predicted state, confidence, explanation, visible clues, and optional warning.
7. The frontend displays that result as the Dogesh AI Report.

One important design decision was to avoid hard-coded responses. Earlier prototype behavior could return the same mood regardless of the uploaded image, so the final version uses Gemini's actual vision analysis to generate the result from the image itself.

Dogesh also avoids forcing an emotional label when there is not enough visual evidence. The model can return "Unclear" instead of inventing an emotion.

Tech Stack

• Next.js
• TypeScript
• React
• Google Gemini / Google AI
• Next.js API Routes
• Tailwind CSS
• Browser image upload APIs

Prize Categories

Best Use of Google AI

Dogesh uses Google Gemini to perform the core image understanding and dog behavior interpretation.

Gemini analyzes the actual uploaded dog image and produces structured information about the dog's visible body language, including the likely emotional state, confidence, supporting clues, and potential warning signs.

#InternationalDogDay #DEVWeekendChallenge #GoogleAI #Dogesh #TypeScript