Age and Gender Identifier
(AGNIFIER)
YASH BHALE (23106088)
LUCKY GUPTA (23106118)
VEDANT AHER (23106123)
ARYAN KOTUR (23106094)
Project Guide
Prof. Priyanka Patil
Outline
• Introduction
• Literature Survey of the Existing
• Systems Limitations of the Existing
• Systems Problem Statement System
• Design
• Technologies and Methodologies
• Implementation Screenshots(Partial )
• Conclusion
• References
Introduction
• Real-Time Observation:
❑ Age and gender detection is widely used in
applications like targeted advertising, security, and
user profiling.
❑ Current solutions often lack accuracy or require complex
setups.
• Motivation:
❑ To create a simple, user-friendly web application for age and
gender detection
• Objectives:
❑ Develop a website that detects age and gender
from uploaded images or camera captures.
❑ Store detection results in a database for future
reference.
❑ Provide an intuitive interface for users
Limitations in existing systems:
❑ Model Loading Issues: There may be instances where
the age and gender detection models fail to load, leading to a
poor user experience.
❑ Error Handling: The application needs robust error
handling to manage issues during image processing, such as
invalid image formats or server errors.
❑ User Interface: The user interface should be intuitive,
allowing users to easily switch between uploading images and
using the camera for real-time detection.
❑Performance Optimization: The application should
ensure that image processing is efficient to provide timely
responses to user actions.
Literature Survey of the Existing System
Research Paper Summary Limitation Adaptation
Age and Gender Develops a CNN-based age Accuracy decreases with Improving accuracy by
Detection [2023] and gender detection system variations in lighting, expanding the dataset,
with high accuracy (92% for expressions, poses, or using multi-task learning,
age, 96% for gender) from underrepresented features in the and fine-tuning pre-trained
facial images. training dataset. models.
Gender and Age Develops a CNN-based age cosmetics, lighting conditions, Model treats age prediction
detection using and gender detection system obstructions, and facial as a classification problem
Deep Learning with high accuracy using a expressions, make it difficult to instead of regression.
[2021] Kaggle dataset. achieve precise predictions.
Age and Gender Facial image processing: Small and imbalanced dataset SENet50_f pre-trained
Prediction using 76.3% age, 86.6% gender reduces accuracy, especially for model improves
Deep CNNs and accuracy. ages above 70. performance via transfer
Transfer Learning learning.
[2021]
Human Age and Facial image processing Lighting conditions impact age Combining edge detection
Gender Estimation method achieves 76.3% age and gender accuracy. and wrinkle density
using Facial Image accuracy and 86.6% gender maintains accuracy across
Processing [2020] accuracy. facial databases.
Problem statement
Users of the New Era struggle because there's no centralized platform for
age and gender detection. This leads to fragmented image processing,
poor handling of uploads and cameras, and communication issues
when errors arise. As a result, users feel disorganized, frustrated, and
dissatisfied with the apps/webapps.
System Design
Technologies and Methodologies
Frontend: HTML , CSS , JavaScript
Libraries: Flask, OpenCV, NumPy, base64
Models: Pre-trained models like Caffe (for face detection) and
custom CNNs for age and gender classification.
Dataset: Publicly available datasets such as Adience, UTKFace, or
IMDB-WIKI for training and testing.
Development Tools: PyCharm, Visual Studio Code(IDE)
Implementation (partial screenshots):
Home page
Implementing camer access and error
detection:
Upload page:
Implementing Upload page with error
detection:
Conclusion
This project demonstrates the feasibility of using OpenCV and Python for
automated age and gender determination. By leveraging pre-trained deep learning
models and robust preprocessing techniques, the system achieves significant
accuracy and efficiency. Further enhancements, such as fine- tuning on diverse
datasets and optimizing for edge devices, can expand the applicability of this
system.
References:
• Welcome to Flask — Flask Documentation (3.1.x)
• OpenCV documentation index
• Base64 – Wikipedia
• HTML Tutorial
• CSS Tutorial
• IJRAR23C2733.pdf
• VanderAalst,W.(2016).DataScienceinAction. Process
Mining, 3– 23.doi:10.1007/978-3-662-49851-4_1
• Y. H. Kwon and N. Da Vitoria Lobo, “Age classification
from facial images,” Computer Vision and Image
Understanding, vol. 74, no. 1, pp. 1–21,1999.
• Used AI (ChatGPT) for code suggestions, corrections,
and implementation ideas.
• Research papers as listed in the literature survey part
of the presentation.
• Used Lucid.app for System Design
Thank
You...!!
