NTU-Corridor Visible-Infrared Person ReID Dataset

The NTU-Corridor dataset was collected by the Rapid-Rich Object Search (ROSE) Lab, Nanyang Technological University (NTU), Singapore. This initiative was the result of a collaboration between the ROSE Lab and the Defence Science and Technology Agency (DSTA) of Singapore.

NTU-Corridor is a Visible-Infrared Person Re-identification (VI-ReID) dataset collected around the outdoor campus grounds of NTU, spanning 136 multi-modality cameras. Those cameras can automatically switch to NIR mode when illuminance is insufficient. Compared with previous datasets, ours provides IR images that are captured under poor illumination environments. Due to the wide area of coverage, the NTU-Corridor dataset consists of both indoor and outdoor scenes with large changes in viewpoint, illumination, and resolution that manifest even within individual wide-angle cameras. It is worth noting that our surveillance cameras are ceiling-mounted, thus our images have the same Top-Down view as application scenarios. There are 128 appearance identities with signed privacy agreements for using their images for academic purposes. To the best of our knowledge, it is the only public VI-ReID dataset with privacy agreements.

 

Sample Images from NTU-Corridor

 

Dataset Statistics

The following table presents a breakdown of the dataset attributes of NTU-Corridor:

  • Property (Value):
  • # Identity (128)
  • # Camera (136)
  • # Images (29783)
  • # Environment (48)
  • IR Type (Near-Infrared)
  • View (Top-Down)
  • Illumination (Variable)
  • Average Image Size (184×456)
  • Colors of clothing (red, black, grey, blue, white, green, purple, pink, multicolor, yellow, orange, brown)
  • Types of clothing (t-shirt, polo-shirt, shirt, dress, jacket)

 

Image File Naming Rules

Given an example image file: "1673_2019-06-25_14602_(female)_CompEngg2-S2-B3a_16_20h19m33s_blue_black_t-shirt_short_267_0.jpg", tokenize the file name using the underscore “_” as separator. The following table describes each token in order of appearance as applied to our example:

  • 1673 (internal record)
  • 2019-06-25 (date)
  • 14602 (person_id)
  • (female) (gender)
  • CompEngg2-S2-B3a (location)
  • 20h19m33s (time)
  • 16 (Internal Record)
  • blue (upper color)
  • black (lower color)
  • t-shirt (upper type)
  • short (lower type)
  • 267 (image index)
  • 0 (internal record)
Request (Account Required)