Sale!
End-to-End CNN-Based System for Human Detection in Fire using a Flask Web App
End-to-End CNN-Based System for Human Detection in Fire using a Flask Web App - Image 2
End-to-End CNN-Based System for Human Detection in Fire using a Flask Web App - Image 3
End-to-End CNN-Based System for Human Detection in Fire using a Flask Web App - Image 4
End-to-End CNN-Based System for Human Detection in Fire using a Flask Web App - Image 5
End-to-End CNN-Based System for Human Detection in Fire using a Flask Web App - Image 6
End-to-End CNN-Based System for Human Detection in Fire using a Flask Web App - Image 7
End-to-End CNN-Based System for Human Detection in Fire using a Flask Web App - Image 8
End-to-End CNN-Based System for Human Detection in Fire using a Flask Web App - Image 9
End-to-End CNN-Based System for Human Detection in Fire using a Flask Web App - Image 10
End-to-End CNN-Based System for Human Detection in Fire using a Flask Web App - Image 11
End-to-End CNN-Based System for Human Detection in Fire using a Flask Web App - Image 12
End-to-End CNN-Based System for Human Detection in Fire using a Flask Web App - Image 13
End-to-End CNN-Based System for Human Detection in Fire using a Flask Web App - Image 14
End-to-End CNN-Based System for Human Detection in Fire using a Flask Web App - Image 15
End-to-End CNN-Based System for Human Detection in Fire using a Flask Web App - Image 16
, ,

End-to-End CNN-Based System for Human Detection in Fire using a Flask Web App

Original price was: ₹2,599.00.Current price is: ₹999.00.

The following contents are Downloadable immediately after the successful payment

  • Source Code of the project

  • Flask Web Application

  • Dataset

  • Trained Model

  • Project Report (Both PDF and Word files)

  • Power Point Presentation (PPT)

  • Instructions to install the necessary Software and Libraries

  • Step-by-step instructions to execute the project (Instructions file)

Project Description

This comprehensive project presents an End-to-End Convolutional Neural Network (CNN) integrated with a Flask web application for detecting humans in fire scenarios using images and video streams. The system combines a sophisticated binary fire classifier model for human detection to identify critical risk situations where individuals are present in fire-affected frames.

The proposed solution implements a fully automated web-based interface that enables:

  • Real-time processing of images and video feeds
  • Frame-by-frame analysis with advanced detection capabilities
  • Live streaming of annotated video outputs with progress monitoring
  • High-confidence predictions for emergency response scenarios

Our system achieves exceptional performance metrics:

  • Test Accuracy: 96.26%
  • Test Loss: 0.1795
  • Precision on Fire Images: 96%
  • Recall on Fire Images: 100%
  • F1-Score: 0.95

The integrated pipeline significantly enhances early fire response capabilities, supports automated safety monitoring systems, and demonstrates remarkable accuracy in recognizing humans in fire environments. This work bridges the gap between academic research and practical deployment by providing a production-ready solution for emergency response teams and safety monitoring systems.

Problem Statement

The challenge is to design and implement an end-to-end system capable of:

  1. Accurately detecting fire in images and video streams with minimal false positives
  2. Identifying human presence within fire-affected areas for immediate risk assessment
  3. Processing video streams in real-time
  4. Providing a user-friendly interface enabling non-technical personnel to operate the system
  5. Supporting emergency response decision-making with confidence scores and clear visualizations

Objectives

  1. Develop a CNN-based fire detection model achieving >95% accuracy with high recall (>95%)
  2. Integrate a YOLO-based human detector for accurate person localization
  3. Design a Flask web application supporting both image and video uploads
  4. Implement real-time video streaming with frame-level annotation
  5. Evaluate system performance on comprehensive test datasets
  6. Create an extensible architecture supporting future multi-class detection
The proposed system consists of six integrated components working in concert:

  1. Data Preprocessing Pipeline: Standardizes dataset preparation, applies augmentation, and ensures consistent normalization across all inputs
  2. CNN Fire Classifier: Performs binary classification (fire/non-fire) with high precision and recall
  3. Human Detector: Localizes persons in images with bounding boxes and confidence scores
  4. Decision Engine: Combines fire and human detection outputs for comprehensive risk assessment
  5. Flask Web Application: Provides user-friendly interface with upload, processing, and result visualization capabilities
  6. Annotation Module: Creates visual representations with color-coded bounding boxes, confidence labels, and scene statistics

Video Demonstration of End-to-End CNN-Based System for Human Detection in Fire Using a Flask Web App

Installing required Software and Running the Project

Installation (One time process)

1. Install the Anaconda Python Package and TensorFlow 2.10

  Follow this video to understand, how to Install Anaconda and TensorFlow 2.10

  Important Note:

  If TensorFlow is not running properly, downgrade the numpy version to 1.26.4 using the below command

  >> pip install numpy==1.26.4

 2. Open anaconda prompt (Search for Anaconda prompt and open)

3. Change the directory to the project folder using the below command

  >> cd path_of_project_folder

Example: Let’s say project is present in D drive

 >>D:

 >>cd D:\Detection-Of-Humans-In-Fire

 4. Activate the virtual environment created while installing TensorFlow using the command

  >> conda activate tf

   Note: Here tf is the virtual environment name created while installing TensorFlow

 5. Now install the required libraries using the below command

  >> pip install -r requirements.txt

 

Follow the steps to train the model after installing the requirements.

1. Open anaconda prompt (Search for Anaconda prompt and open)

2. Change the directory to the project folder using the below command

  >> cd path_of_project_folder

Example: Let’s say project is present in D drive

 >>D:

 >>cd D:\Detection-Of-Humans-In-Fire

 3. Activate the virtual environment created while installing TensorFlow using the command

  >> conda activate tf

4. Next to train the model open the Jupyter Notebook using the below command

  >> jupyter notebook

 5. Open the End-to-End-CNN-for-Detection-of-Humans-in-Fire-with-Flask-Web-App.ipynb and run all cells

6. Once the training is completed the trained model that is fire_detection_model.h5 will be stored in the models directory

Follow the steps to run the Flask Application after installing the requirements and Training the Model.

1. Open anaconda prompt (Search for Anaconda prompt and open)

2. Change the directory to the project folder using the below command

  >> cd path_of_project_folder

Example: Let’s say project is present in D drive

 >>D:

 >>cd D:\Detection-Of-Humans-In-Fire

 3. Activate the virtual environment created while installing TensorFlow using the command

  >> conda activate tf

4. To run the Flask app  type the following command

  >> python app.py

Software Requirements (for Web Application & Backend)

1. Operating System

  • Windows 10 / Windows 11 (64-bit)

  • Ubuntu 20.04 / 22.04 (Recommended for deployment & GPU support)

2. Programming Environment

  • Python 3.9 (supports TensorFlow 2.10 )

3. Backend Software / Frameworks

  • Flask 3.1.x – Web server framework for handling image/video uploads and API responses

  • Werkzeug – Secure file handling in Flask

  • Threading module – For background video processing jobs

  • Jinja2 templates – For rendering front-end HTML pages

4. Machine Learning Libraries

  • TensorFlow 2.10.0 – Used to load & infer the custom Fire Detection CNN model

  • OpenCV (cv2) –

    • Image preprocessing

    • Video frame extraction

    • YOLO inference

    • Annotation of bounding boxes

  • NumPy 1.26.x – Mathematical array operations

  • Pillow (PIL) – Reading uploaded images from the user’s browser

  • imutils 0.5.4 – Efficient resizing & frame utilities for video processing

5. Detection Models Required

  • fire_detection_model.h5 – Custom-trained CNN for fire detection

  • yolov3.cfg – YOLOv3 model configuration

  • yolov3.weights – Pre-trained YOLOv3 weights

  • coco.names – Class labels (must include “person”)

Hardware Requirements

Minimum Requirements (for running locally or small-scale deployment)

Suitable for testing or low-resolution inputs.

  • CPU: Dual-core 2.0 GHz (Intel i3 / Ryzen 3)

  • RAM: 8 GB

  • Storage: 10 GB free (models + processed videos + logs)

  • GPU: Not required (CPU inference works but slower)

  • Camera (optional): Basic webcam for future real-time extension

Happy Learning

Still need help to set up and execute the project

  • Setup and modification are paid services based on requirements.
Categories: , , Tag:

Reviews

There are no reviews yet.

Be the first to review “End-to-End CNN-Based System for Human Detection in Fire using a Flask Web App”

Your email address will not be published. Required fields are marked *

Related products

Welcome to VTUPulse.com

Download Final Year Project -> Click Here

This will close in 12 seconds