Cars Dataset (Positive and Negative)

The dataset is organized into 2 folders (positive and negative) which contains total 2,800 images. The data is split into 1,230 positive images (car images) and 1,570 negative images (background images).

The dataset is arranged in such a way that HAAR Cascade Classifier training can be done with it. This structure is described in OpenCV docs (https://docs.opencv.org/3.4/dc/d88/tutorial_traincascade.html).

Positive Directory structure:
'img' folder witch contains all positive images and 'info.dat' file which contains all positive image descriptions. The first element of the line is the filename, followed by the number of object annotations, followed by numbers describing the coordinates of the objects bounding rectangles (x, y, width, height).

/img
img1.jpg
img2.jpg
. . . .
info.dat

Inside the file info.dat:
img/img1.jpg 1 140 100 45 45
img/img2.jpg 2 100 200 50 50 50 30 25 25

Negative Directory structure:
'img' folder witch contains all negative images and 'bg.txt' file witch contains all negative image directory list.

/img
-img1.jpg
-img2.jpg
. . . .
bg.txt

Inside the file bg.txt:
img/img1.jpg
img/img2.jpg
. . . .