🖼️ Image Search App (Google Cloud Vision + Google Cloud Vertex AI AutoML model + Gemini API)

This application allows users to:

• Upload images and detect objects using Google Cloud Vision, Vertex AI AutoML, and Gemini API.
• Search images using tags (object labels) saved in Firestore from above results.
• Upload videos and perform label detection using Google Video Intelligence API.

---

🔧 Features

1. Image Upload & Object Detection

   • Vision API: Object localization
   • Vertex AI: AutoML object detection
   • Gemini: Multimodal vision using Gemini 2.0

2. Image Search

   • Search previously uploaded images using tag-based metadata from detections.

3. Video Analysis

   • Upload a video and detect labels using Google Cloud's Video Intelligence API.

---

🚀 Pre-requisites

Before running the app, make sure to set up the following:

1. ✅ Enable APIs in GCP:

   • Vertex AI API
   • Cloud Vision API
   • Video Intelligence API

2. 🔑 Create Gemini API key from:

   • Google AI studio

3. 🔗 Link GCP project to Firebase:

   • Use Firebase Console to associate your GCP project.

4. 🛠️ Install Google Cloud SDK

5. 🧠 Upload and train your model in Vertex AI:

   • Vertex AI Training Guide
   • PASCAL VOC 2008 dataset: https://www.kaggle.com/datasets/sulaimannadeem/pascal-voc-2008

6. 🪣 Create a Cloud Storage bucket (used for image storage)

7. 🔐 Create and update Secret Manager entries for:

   • GCS_BUCKET_NAME
   • VERTEX_PROJECT_ID
   • VERTEX_ENDPOINT_ID
   • VERTEX_LOCATION (e.g., us-central1)
   • GEMINI_API_KEY

8. 📄 Update PROJECT_ID in app.yaml as your GCP project number

9. 🔥 Create Firestore database (used to store image metadata and search results)

   • Create a default database in GCP Firestore: Ensure it is set up with a native structure.

10. Enable App Engine in GCP. Make sure the service account that is created has Secret Manager Admin and Storage Admin roles.

---

🛠️ Deployment Steps

# In Google Cloud SDK shell
gcloud init                    # Select the appropriate Google account and project
gcloud app create              # Choose the region where the app will be deployed
# In VS Code, go to root directory
gcloud app deploy app.yaml --quiet  # Deploy the app
gcloud app browse              # Open the deployed application URL