Pattern Recognition System sem. 1, 2015-2016

Monday - 11-17, room 106 Dorobantilor

Attendance and Marks


Laboratory topics

  1. RANSAC: fitting a line to a set of points:
    prs_lab_01e.pdf (English);
    prs_lab_01r.pdf (Romanian);
    line.bmp ; line-res.bmp
  2. RANSAC: fitting a circle to a set of points:
    prs_lab_02e.pdf ;
    prs_lab_02r.pdf ;
    circle.bmp ; circle-res.bmp
  3. Hough Transform for line detection
    prs_lab_03e.pdf ;
    prs_lab_03r.pdf ; ;
  4. Pattern Matching Using Distance Transform
    prs_lab_04e.pdf ;
    prs_lab_04r.pdf ; ;
  5. Histograms of Oriented Gradients
    prs_lab_05r_hog.pdf; ;
  6. Statistical Data Analysis
    lab_06r_statisticDA.pdf; ;
    background_white.bmp ;
  7. Density Estimation
    prs_07_points.txt ;
    prs_lab_07e_background_white.bmp ;
  8. Naive Bayes Classifier: Simple Digit Recognition Application
    prs_lab_08e_bayes.pdf ;
    DIGIT_imgs.ZIP ; TEST_imgs.ZIP ;
  9. Linear Classifiers: The perceptron classifier prs_lab_09e_perceptron.pdf;
    lab_09r_perceptron.pdf; ;
  10. Linear Discriminant Analysis prs_lab_10e_lda.pdf; ;
  11. Ensemble methods: AdaBoost classifier prs_lab_11e_adaboost.pdf;
    lab_11r_adaboost.pdf; ;
  12. SVM Classification lab_12e.pdf;
    TEST_svm.ZIP ;
    SVM_images.ZIP ;
  13. Classifier Evaluation 14_EvaluareClasificatori.pdf ;


Possible topics:
  1. Pedestrian detection in intensity images by means of simple classifiers (Bayesian, Perceptron, AdaBoost)
  2. Gesture recognition by means of a Bayesian approach
  3. Face detection in grayscale / color images
  4. Signature recognition
  5. Pedestrian detection in thermal images by means of simple classifiers (Bayesian, Perceptron, AdaBoost)
  6. Handwritten character recognition
  7. Face recognition in grayscale / color images
  8. Pedestrian detection in intensity images by means of hierarchical template matching.
  9. Pedestrian detection in thermal images by means of hierarchical template matching.
  10. Traffic sign detection and recognition in color images.
  11. Coin recognition in color images.
  12. Banknote recognition in color images.
  13. Indoor vs outdoor scene classification.
  14. Image segmentation of landscape and urban scenes.
Note - for the implementation you can use the OpenCV computer vision library. On request a sample demo application will be provided.

Project evaluation

The final mark of the project will be formed by combining individual marks for the following evaluation items. The weights of each evaluation item are given below.
  1. Functionality - 30%
  2. Documentation - 10%
  3. Code structure - 10%
    • Functionality, documentation and code structure will be evaluated by the teacher in the next 7 days after you present the project.
  4. Semester activity - 40%
  5. Final presentation - 10% (includes running the project and presenting the documentation during the project presentation session
You will create an archive having the form: Where The archive will contain:
  1. A folder named src where you will put the code of the project.
  2. A text file named README where you will write how the program is to be run, and what special constrains and dependencies are needed for running the project. (Eg. Write if you have used Diblook, or OpenCV or if there are any images needed for running the project - for example datasets containing cars, or if you have any configuration files that you have used etc).
  3. A folder named doc that will contain the Documentation (.doc or .pdf or .ppt)
  4. Optionally - a folder named resources where you will place other resources needed for running the project (for example images, libraries etc). If you have used openCV just mention this in the README file, there is no need to copy it in the resources directory.