CMPE180B Group Project - Inventory Management System with Image Recognition and MongoDB Backend

Group Name: DBLegends
Group Members:

• Lu Li
• Bhavya Jain
• Apoorva Adimulam
• Nivedita Nair

---

Introduction

The Inventory Management System leverages machine learning, Flask for the user interface, LLM-based image recognition, and MongoDB as a NoSQL database. This system processes images to identify products, manages inventory data, and provides comprehensive CRUD operations for inventory management.

---

Project Structure

Application Working Directory: integrate_image_recog_backend_mongodb

Files:

1. app.py

   • Main Flask application file.
   • Handles routes for processing images, viewing inventory, updating inventory, deleting products, and uploading JSON files.
   • Contains MongoDB functions for inserting, updating, and deleting products.
   • Uses the ProductImageProcessor class for image processing.

2. create_db_with_json.py

   • Script to create and populate the MongoDB database using a JSON file.
   • Drops the existing database and inserts data from products.json.
   • Ensures unique indexing on HashID.
   • Creates initial batches for every unique product.

3. csv_json_conversion.py

   • Converts inventory.csv into a JSON file named products.json.
   • Generates HashID for each product based on its name and brand.

4. insert_update_from_image.py

   • Contains the ProductImageProcessor class for processing images.
   • Defines MongoDB functions for generating HashID and inserting/updating products.
   • Processes images to identify products and update the database.
   • Adds new batches to existing products or inserts new records for products not found in the database.
   • Supports timestamp-based batch tracking.

5. delete_op.py

   • Contains the delete_product function for deleting records from the database using a combination of product and type.

6. range_query.py

   • Script to perform range queries on the MongoDB database.
   • Creates an index on the Product field for faster queries.
   • Retrieves all items for a specific product.

7. inquiry.py

   • Script to query the MongoDB database for a specific product.
   • Generates HashID and retrieves product data based on the hash.

8. dependencies/setup_mongodb_pymongo.sh

   • Bash script to set up MongoDB and pymongo on a Linux system.
   • Installs MongoDB, starts the service, and installs pymongo.

9. dependencies/LLM_dependencies.sh

   • Bash script to set up LLM dependencies on a Linux system.
   • Installs transformers, torch (GPU support version), and einops.

10. dependencies/frontend.sh

    • Bash script to set up frontend dependencies.
    • Installs Flask.

---

Project Setup

Prerequisites

• Python 3.x
• MongoDB
• Flask
• pymongo library
• transformers library
• Pillow (PIL) library

Installation

1. Clone the repository:

   git clone https://github.com/luliCloud/CMPE180B_Group-project.git
   cd CMPE180B_group-project/integrate_image_recog_backend_mongodb

2. Set up dependencies:

   chmod +x dependencies/setup_mongodb_pymongo.sh
   ./dependencies/setup_mongodb_pymongo.sh
   
   chmod +x dependencies/frontend.sh
   ./dependencies/frontend.sh
   
   chmod +x dependencies/LLM_dependencies.sh
   ./dependencies/LLM_dependencies.sh

---

Tests

Rigorous tests have been written for all key components of the inventory management system.

Directory: tests

Files:

1. test_app.py

   • Tests all functions in app.py, including database creation and CRUD operations.

2. test_db_creation.py

   • Tests all functions in create_db_with_json.py.
   • Verifies normal database creation.

3. test_db_delete.py

   • Tests the delete_product function in delete_op.py.
   • Verifies records can be deleted using product and type.

4. test_image_process.py

   • Tests image processing functions in insert_update_from_image.py.
   • Verifies uploaded images can be found in the target directory, and their information is properly recognized and extracted by LLM.

5. test_insert_update.py

   • Tests insert and update functions in insert_update_from_image.py.
   • Verifies extracted information is inserted or updated in the database accordingly.

6. test_query_range_index.py

   • Tests range_query.py and inquiry.py.
   • Verifies second indexes can be created, and both precise and range queries work as expected.

---

Execution Instructions

Run with the UI

1. Start the Flask application:

   python app.py

2. Access the application in a web browser:

   http://localhost:5000
   

Run with the Backend

1. Create the database using a JSON file:

   python3 create_db_with_json.py

2. Insert and update information from newly scanned images:

   python3 insert_update_from_image.py

3. Delete records from the database:

   python3 delete_op.py

4. Perform precise queries:

   python3 inquiry.py

5. Perform range queries:

   python3 range_query.py

6. Create second indexes:

   python3 range_query.py