Part I: Data Collection
1- Dataset collection: Collect hundreds of grayscale images (200 hundred is the minimum). The dataset should include a least 4 categories. (e.g cars, horses, buildings, tigers…). The categories should be balanced (approximately the same number of images for each category).
2- Dataset labeling:
a. Assign a Label to each image. (E.g. For car image assign 1, horse 2, building 3 and tiger 4…).
b. After ordering your images, construct a vector containing the label of each image in same order of the images.
c. Save all images in a directory called ImagesDatayourName
d. Save the list of the image names ordred and its corresponding labeling vector in [url removed, login to view] and [url removed, login to view] files.
3- If needed, enhance or/and restore the images. Explain why it is need and the choice of the performed method.
Part II: Feature Extraction
1- For each image from the image Data, extract at least one color feature.
2- Normalize you data. For each Feature, make each entry between 0 and 1 for all images.
3- Save each Feature in an MxN matrix where M is the number of images and N is the dimension of your Feature vector. You have to use the same order of the images as the one in your name and label files.
4- Write a Readme File that describes each Feature used, indicates the corresponding name of the Feature file, and the name of the Matlab function that extract it.
Part III: Retrieval
1. Query Image Q (Selection and display): Select and display the query image.
2. Retrieved image Rank 1, 2,3 and 4: Display the resulting retrieved image.
3. Dist(Q,R1), Dist(Q,R2), Dist(Q,R3), Dist(Q,R4): display the distance between the query and the retrieved image under the retrieved one.
4. Score of this retrieval: evaluates the current retrieval.
Computer Science Department
College of Computer and Information Sciences
King Saud University
The final report should roughly have the following format:
Introduction - Motivation
o Details of the experiments; observations
Conclusions (draw your own conclusions. Be creative)
You should email your final report to the instructor. A hard copy submission is not required.
Once the project is completed, the following is expected of you:
a. A demonstration of your project where you show the features of your system, such as its correctness, efficiency, etc. You should be prepared to answer detailed questions on the system design and implementation during this demo. We will also examine your code to check for code quality, code documentation, etc.
b. You should also hand in a completed project report which contains details about your project, such as main data structures, main components of the algorithm, design of the user-interface for input/output, experimental results, e.g. charts of running time versus input size, etc.
c. You should also turn in your code and associated documentation (e.g. README files) so that everything can be backed up for future reference.
d. Email your code and all associated files to the tutor with “CSC 478-Project<Lastname>” as subject.
Questions and Office Hours
Your instructor is willing to answer your questions about image processing techniques or the experiments. He will not answer questions about coding errors as it is my feeling that, at this point, writing error-free code is your responsibility.
*attached images need to transform to grayscale at the beginning
*Images collectins here:
[url removed, login to view]!aJcw1aKb!WshRnDB9RtSgl412E0OGh6KyFXX3GQVZdakVpJYbjqA
13 freelancers estão ofertando em média $302 para este trabalho
I have already done this kind of project in my machine learning course. It was dogs and cats classification using convolution neural network(Deep learning) results were 74.5%.