Hyunggic !



Vehicle DPM


Introduction
This vehicle DPM was designed and trained especially for autototive active safety applications such as FCW (Forward Collision Warning).

We collected vehicle dataset in various places and weather conditions. It consists of three training sets (set00 ~ set02) and two testing sets (set03 ~ set04), each with 1 or 2 seq files, along with associated annotation files. Due to our purpose, most of samples corresponds to rear view samples of vehicles such as sedan, SUV, truck, and bus. The dataset was very carefully collected based on our investigation about vehicle accident statistics [1]. So, we believe that this dataset could be very valuable for those safety applications.

[1] National Motor Vehicle Crash Casusation Survey
NHTSA, 2008

[Typical Samples]


[Typical Samples with annotations]


[Dataset Statistics]

Left pie chart shows the distribution of vehicle samples and right hand figure shows the distribution of bottom lines of all (i.e., 14,453) bounding boxes. Key numbers are:

Sedan_rear: 5,976
SUV_rear: 4,344
Truck_rear: 390
Bus_rear: 299


Top row shows a height distribution and bottom row shows ratio distribution of each sub-type.




[Trained Sedan (5,976)) / SUV (4,344) Model]


[Trained Truck (390)) / Bus (299) Model]



[Result Video]
Vehicle detection result on Pittsburgh Dataset