Object Detection using Transfer Learning Test phase, srs, design phase and source code final deliverable

Get help with
Phd, Masters Thesis & Mcs Final Project
MBA Final Project
Cs619,Fin619,Mgt619,Bnk619,Hrm619,Mkt619
WhatsApp: 0092-3125718857
Skype: trust_aware
Email: projecthelp77@gmail.com
Click here to Join Our Facebook Page
Click here to Join Our YouTube Channel!

Object Detection using Transfer Learning Test phase, srs, design phase and source code final deliverable

Category

Deep Learning

Abstract / Introduction

The aim of this project is to develop an Object Detection system that can detect the objects in an image. Whenever an object is detected in an image, a rectangular bounding box is drawn around the object. In this Project we will use a dataset of images and train an algorithm which detects different objects in images for example whether people do wear a mask, do not wear a mask, or whether they wear the mask incorrectly. The main advantage of this application is that it helps to develop a detection system/recognition system like a face detection machine which is now commonly used in most organizations to make sure that only authorized person/employee can get entry into the building. Such machines can also be used as attendance marking machines which surely saves time a lot.

Functional Requirements:

1.Reprocessing of provided data set

1. In pre-processing, cleaning of data set is the key process (if required).
2. Dataset: You can download both type of cards(CNIC data available will be provided to enrolled students) from google images. For example you can download 20 for each (you can download more for better accuracy).
3. Annotations: You can use any annotation software (labellimg, labelme, labelbox, VIA (VGG Image Annotator) to annotate your images.
4. Augmentation: after annotations, you can use some augmentation techniques to create more images. For example you can rotate an image every 4 degrees and in this way you can have 90 images from one image. You can use other augmentation approaches too (e.g., even change name, dob etc on cards).
5. Transfer learning: You will train the model on your own dataset. Basically you will use transfer learning and for that download the coco pretrained weights and you can use ResNet50, or any other better architecture (for details refer to the link above).
6. Results Required: You need to plot training and validation curves for 50 epochs. Moreover you need to report the testing accuracy.

Tools:

Software Requirements:

• Operating System: Window 7 and above
• RAM 4 GB or more
• Anaconda OR       jupyter notebook     OR Google Colab (Python)

Download sources: https://anaconda.org/ https://jupyter.org/

Language of the Project:

Python

Note: You can write the Names of Functions of your own choice. Do not use random datasets.

Dataset will be provided through email to the enrolled students.

Supervisor:

Name: Shafaq Nisar