LP-Night dataset containing photos with license plates captured at night.

LP Night Dataset

Dataset Description

This dataset was introduced in the paper “A new augmentation-based method for text detection in night and day license plate images” and “A new U-net based license plate enhancement model in night and day images”. It includes images of different vehicles, blur, multi-script, imbalanced illumination, orientation, and perspective. The images have a resolution {Min: (2067 x 2066); Max: (3120 x 4160) } and are taken at night.

Dataset Samples

Sample AMLPR dataset

Annotation Format

The dataset folder contains images with filenames as IMG_XXXXXXXX_XXXXXX.jpg and annotations as IMG_XXXXXXXX_XXXXXX.xml. The xml files look like the following:

<annotation>
	<folder>lp_data_100</folder>
	<filename>IMG_XXXXXXXX_XXXXXX.jpg</filename>
	<path>/home/pinaki/work/acpr_conf/training_data/lp_data_100/IMG_XXXXXXXX_XXXXXX.jpg.jpg</path>
	<source>
		<database>Unknown</database>
	</source>
	<size>
		<width>4160</width>
		<height>3120</height>
		<depth>3</depth>
	</size>
	<segmented>0</segmented>
	<object>
		<name>LONDON</name>
		<pose>Unspecified</pose>
		<truncated>0</truncated>
		<difficult>0</difficult>
		<bndbox>
			<xmin>1119</xmin>
			<ymin>751</ymin>
			<xmax>1430</xmax>
			<ymax>898</ymax>
		</bndbox>
	</object>
  ...
  ...
  <object>
		<name>...</name>
		<pose>Unspecified</pose>
		<truncated>0</truncated>
		<difficult>0</difficult>
		<bndbox>
			<xmin>XXXX</xmin>
			<ymin>XXXX</ymin>
			<xmax>XXXX</xmax>
			<ymax>XXXX</ymax>
		</bndbox>
	</object>
</annotation>

If we are unable to read the text region (i.e., it is not legible), then we use <name>###</name> in the xml fields.

License / Terms of Use

The annotations are released under Creative Commons Attribution 4.0 License. The authors do not claim rights to the photos in the LP-Night dataset. We only collect publicly available images and annotate them for research purpose. Request to exclude any/all photos from LP-Night dataset should be addressed to mail@pinakinathc.me

How to Cite this dataset

Chowdhury et al. "A new U-net based license plate enhancement model in night and day images", In Asian Conference on Pattern Recognition, 2019.

Chowdhury et al. "A new augmentation-based method for text detection in night and day license plate images", Multimedia Tools and Applications, 2020.

or

@inproceedings{chowdhuryACPR2019,
  title={A new U-net based license plate enhancement model in night and day images},
  author={Chowdhury, Pinaki Nath and Shivakumara, Palaiahnakote and Raghavendra, Ramachandra and Pal, Umapada and Lu, Tong and Blumenstein, Michael},
  proceedings={Asian Conference on Pattern Recognition},
  year={2019}
}

@article{chowdhuryMTAP2020,
  title={A new augmentation-based method for text detection in night and day license plate images},
  author={Chowdhury, Pinaki Nath and Shivakumara, Palaiahnakote and Pal, Umapada and Lu, Tong and Blumenstein, Michael},
  journal={Multimedia Tools and Applications (MTAP)},
  year={2020}
}

Download this dataset

Please note, downloading this dataset means you agree to the above mentioned Terms and Conditions. This dataset is only provided for non-comercial purposes. For comercial use, please contact authors at mail@pinakinathc.me

Download from Google Drive: [Raw Unlabelled] [Annotated 100 images]