Blanking out images in an object

Please write a report on your final project. Your report should include the following

1. Title of the Project
2. Summary: summarize the report in 3-5 sentences
3. Introduction: why the problem your project is trying to solve is important and/or how it can be used in real life
4. Methodology: what methodology was used to solve the problem
5. Experiments: show the results (output images) of experiments you conducted. You may want to compare your output images with the input image(s), and the output image(s) for other methodologies you chose to compare with
6. Conclusion(s): summarize the conclusion(s) of your experiments
7. References: list at least three references of articles in the literature that is related to what you propose to do.*

Requirements:

1. Length of the report: no more than 5 pages (not including references).
2. Formatting: Single space. Font size: 12 point. Margins: 1 inch on each side

Must contain all the sections/information above.

Standard:

1. Complete: can not miss any one of Sections No. 2~No. 7 above
2. be free of grammar and/or spelling errors.
3. demonstrate the knowledge and understanding of the content material of this course.
4. Literature search beyond the textbook.

Submission:

Upload the following documents on blackboard, under the posting of this assignment. (Do not zip)

1. your report in MS Word format
2. Your source code files (Your code must be commented)
3. Your input image (s), except for the ones that are the sample images in the book

Term Project Proposal: Blanking out objects in an image
Summary
The goal of this project is to develop an efficient selection process from digital images while
maintaining image quality and aesthetics. I will explore image preprocessing, object detection,
and object removal processes. The goal is to provide users with both automated and manual
removal methods through user-friendly interfaces. The project will contribute to the development
of digital image processing and will have applications in a variety of industries, including image
editing.
Problem Statement
The problem I aim to achieve is to select objects from digital images and ensure that the edited
images maintain both quality and visual cohesion This task is difficult because of the need to
detect and remove objects without introducing visual artifacts or compromising the aesthetics of
the image.
Related Works
Previous research has investigated feature extraction from digital images using various methods.
For example, Redmon et al. (2016) proposed YOLO (You Only Look Once), a real-time object
detection system that can be adapted for object removal. Additionally, techniques such as
visualization and filling have been used to remove objects. These methods form the basis of our
project’s methodology.
Methodology
My methodology is as follows:
– Processing the image: enhancing image quality through denoising, contrast enhancement,
and image segmentation.
– Detecting the object
– Removal of the object: implement algorithms for selective object removal while
maintaining image aesthetics.
– User Interface: making sure it is user friendly for other people to upload and process their
images, including options for automatic and manual object removal.
Design of the experiment
The experiment will consist of testing the system on a range of datasets and test cases to evaluate
its performance. I plan to use a minimum of three different datasets/test cases to assess the
system’s robustness across various scenarios. The datasets will include images with diverse
objects of interest. I will evaluate the system’s accuracy by comparing the results with ground
truth data and assess image quality by considering visual artifacts and overall aesthetics.
References
Redmon, J., Divvala, S., Girshick, R., & Farhadi, A. (2016). You Only Look Once: Unified,
Real-Time Object Detection. arXiv preprint arXiv:1506.02640.
Ren, S., He, K., Girshick, R., & Sun, J. (2017). Faster R-CNN: Towards Real-Time Object
Detection with Region Proposal Networks. IEEE Transactions on Pattern Analysis and Machine
Intelligence, 39(6), 1137-1149.