A dataset for vehicle classification in darkness collected and labeled in the Autonomous Robots Lab at the University of Nevada, Reno.
By Jack Currie, Brenda Penn, and Dusty Barnes
The data in gt10913.txt is organized as follows, with single spaces delimiting each of the values on a line. The n-th row in the gt10913.txt file corresponds to the n-th image in the data directory, and is indicated by the "Image Number" field in each row.
The "x" and the "y" are the x-y coordinates of the top left corner of the vehicle with relation to the top left corner of the image.
0 4 [139 248 163 80 ] [272 230 132 49] [382 219 66 36] [425 208 49 39]
This dataset contains 10913 gray-scale images of night-time images of roads labeled img_0.jpg --> img_10913.jpg. The images all have dimensions of 1280 x 1024 pixels (width x height). Not all images in the dataset contain vehicles, though the majority (~90%) do. The data was collected using a PointGrey Chameleon 3 Grayscale camera (CM3-U3-13S2C-CS-BD).
Some images were collected from a moving bus, and others were collected with roadside cameras (having static backgrounds).
0 - 2006 | Filmed from Bus, non-static background | Dynamic
2007 - 8247 | Filmed from roadside at intersection of N. Virginia St. and College Dr. (3 different vantage points) | Static
8248 - 10193 | Filmed from roadside at intersection of Sierra St. and College Ct. (2 different vantage points) | Static
Indicative images are shown below:
You may download the complete dataset following this link or simply clone this repository.
Another similar nighttime dataset from Long-Chen from Sun Yat-sen University can be found following this link.
This dataset provides an additional ~6000 images which are all in color, and will certainly be of interest to you if you are investigating this dataset.
You can contact us for any question or remark:
This material is based upon work related to the Intelligent Mobility project supported by the Governor's Office of Economic Development.