MIO-TCD

The dataset consists of total 786,702 images with 648,959 in the classification dataset and 137,743 in the localization dataset acquired at different times of the day and different periods of the year by thousands of traffic cameras deployed all over Canada and the United States. Those images have been selected to cover a wide range of challenges and are representative of typical visual data captured today in urban traffic scenarios. Each moving object has been carefully identified by nearly 200 persons to enable a quantitative comparison and ranking of various algorithms. This dataset aims to provide a rigorous benchmarking facility for training and testing existing and new algorithms for the classification and localization of moving vehicles in traffic scenes
The dataset is divided in two parts : the “classification challenge dataset” and the “localization challenge dataset”.
NOTE! If you intend to use any part of the MIO-TCD dataset, please cite the following paper :

Z. Luo, F.B.Charron, C.Lemaire, J.Konrad, S.Li, A.Mishra, A. Achkar, J. Eichel, P-M Jodoin
MIO-TCD: A new benchmark dataset for vehicle classification and localization
in press at IEEE Transactions on Image Processing, 2018

Contains 137,743 high-resolution images containing one (or more) foreground object(s) with one of the following 11 labels:

Articulated truck
Bicycle
Bus
Car
Motorcycle
Motorized vehicle (i.e. Vehicles that are too small to be labeled into a specific category)
Non-motorized vehicle
Pedestrian
Pickup truck
Single unit truck
Work van

The goal of this challenge is to correctly localize and classify each foreground object.
Click here to download the python code for computing training error metrics as well as loading and saving bounding boxes.

Participants are free to upload results for the classification AND/OR localization challenge.