Instructions for Final Project Reports

• Please submit your final project report to EEE, plus supporting code
• Reports should be a minimum of 5 pages and a maximum of 10 pages in length in Word or PDF format, called finalreport.doc or finalreport.pdf.
• Please also upload all of the code you developed for the final project (as a Zip file called code.zip), including a README file that briefly explains the functionality of each function that you wrote or use. Your code should be clearly documented, to the extent that someone such as the TA or a professor can understand from the documentation how to run your code, e.g., what types of input images to provide, what output to expect, etc. Include any MATLAB functions are not part of the basic MATLAB toolbox, but that are called by your project functions, e.g., function modules that you developed for earlier assignments.
• Your code should also include a single MATLAB file called demoscript.m that provides an example of how your code runs, e.g., reads in a test image, extracts features in real-time, and uses a classification model or face detection algorithm, etc) that was already learned by your algorithm "offline" to classify or process this test image, and then displays the results. IMPORTANT:
  
• your script should run without requiring any input - so define some variables at the top of the script that define your classifier (e.g., stored in a .mat file to be read in) and your test data.
  
• Also, your demo_script should not take any longer than 1 minute to run!
• If you are using the "default" data sets, you do not need to upload any data. If you are using a non-standard data set, please upload your data (in MATLAB format if possible) in an additional Zip file called data.zip.  If youe data file is too large to upload, please make the data available somewhere on a Web page and upload a text file called "README_data" that points to the location of the data.
• The total size of whatever you upload in total (after compression with Zip) can be as large as 50Mbytes (but no larger).

What your Project Report must contain:

1. Task Description: A brief description of the specific classification or recognition task you addressed.
   
2. Data Sets:  A brief description of what data sets you used in your project. Clearly indicate number of images, resolution, and any other relevant aspects.
   
3. Feature Calculation: The specific image representation (features) you investigated for solving this problem, e.g., provide descriptions of how the features were calculated (if its complicated, you can put pseudocode in an Appendix, for example) and provide examples (e.g., sample images, graphs, etc) if possible.. It should be clear from your description how your features are defined and how they were used for in classification (e.g., if you use templates as features, be very clear on how the template results are used as features for classification).
4. Classification Algorithms: Describe briefly the classification algorithms you used, and in particular any settings that you used (e.g., how was k chosen for kNN?) or any modifications you made to the basic algorithm.
5. Experimental  Setup: A description of how your experiments were conducted. Clearly describe how you ran your experiments so that a reader could repeat the general method you used if they wanted to..
   
6. Experimental Results: Tables, graphs, figures, that present your experimental results. Again there should be enough detail here that a reader can evaluate your project, but avoid having pages and pages of numbers, i.e., the reader should be able to easily see your main results.
   
7. Assessment and Evaluation: Provide a discussion on your evaluation and assessment of the results, e.g., having completed the project do you think the task was difficult or easy for a computer algorithm to try to solve? what were the limitations of your approach? what were the successes? how might you build a better system if you had more time? what aspects of the data set limited your results? And so forth. Be honest in your assessment of your results, e.g., if your accuracies were low you should honestly and objectively discuss why the accuracy may be low. Similarly if the accuracy is high, you should explain why if possible. You can full points if you write a well-written comprehensive report, even if your results are not very accurate: conversely you will get a low score if you write a poor report, even if you have very accurate results.
8. Appendix: Software.  List the main MATLAB functions (plus code in any other languages) that you wrote for the project. Clearly indicate whether each function is either
• written entirely from "scratch" by you for the project
• a modified version of a function you wrote earlier for an assignment
• a copy of a function provided by the instructor (or a modified version of the instructor's code)
• code obtained from the Web or some other similar source.

Grading of the Final Report

• The final project report is worth 35% of your total grade. So please make sure to allow enough time to write a good final project report!  You should write this with a general reader in mind, e.g., another computer scientist who did not take this class - so the report should be fully self-contained and clearly explain where the data came from, what functions you used (whether you wrote them or someone else) and so forth.
• Reports will be graded on
• clear writing style: is the report clear and easy to follow?
• is the report systematic and thorough?
  
• use of figures, graphs, tables, etc, to illustrate the results. Feel free to use color in your figures if it helps explain or clarify a particular point.
  
• technical correctness - e.g., is it clear that you understand the techniques you are using, and are you able to draw appropriate conclusions from your results.
  
• degree of difficulty of the project attempted
• Note: it is most important that you write a clear report and document objectively what methods you implemented and the experimental results you obtained. How well the method worked is not nearly as important as clearly documenting what you did and what you learned from your project.