alexeigith / cvfinalproject

TrackPlayers

TrackPlayers project is the final version of our code to detect players.

It automatically processes an input image, creates a mask to hide a crowd area around the rink, makes players on a rink more distinguishable. After that it passes that image to a HOG detector and saves an output image with rectangles around detected player

cvfinalproject's Issues

Write code to read cvc files

Write a class/method to read csv files (files will contain video file names) and store the result in a vector of strings.
This class/method must be have configurable delimiter and an input file name.

Is point inside rect

Write a function that accepts coordinates of a point and a rectangle, and checks if the point is inside that rectangle.

Write a second method that will check if the point is exactly in the middle of the rectangle accepting a standard error parameter.

(For example, Check if Point(11,10) is in the middle of Rect(20,20) having an error of 2pixels. In this case Point is not exactly in the middle but a 2px error allows us to say that it is)

Detect players

Write a code that will detect players on the rink.

1. Detect player and draw bounding rectangles
2. Save results as a csv file including file name (image name) and all rectangles detected on that image (rectangle properties)

Class/Method that will cut 5 seconds from a video file using opencv

1. Accept a file name
2. Determin the total duration of the video (in seconds?)
3. Randomly pick a start point (Must have at least 5 seconds of video after that. For example if the video is 60 sec long, the start point cannot be 56 sec or more).
4. Cut 5 seconds from the file starting at a start point calculated at step 3)
5. Save file using the same name and adding a prefix "-short". (Example: file "1.mp4" will be trimmed to "1-short.mp4")

Create a program that can be used for annotation

1. Read image/image list
2. Display an image (one by one if many)
3. Put a circle at the point where the user clicked a mouse, ask if that point is okay.
4. Save all the coordinates of points to the array
5. Write all the selected for current image to a file
6. Repeat for all files, appending coordinates and file names to the end of a csv file

A program to test performance of the player detection algorithm

For this program we'll use code for player detection, and a set of points created using annotation tool.
We will be comparing how many players the program has detected by checking if points annotated manually are inside those rectangles that represent players.