The Street View Text Dataset - Revision history

Dimos at 11:40, 27 October 2012

2012-10-27T11:40:24Z

Dimos at 13:24, 16 October 2012

2012-10-16T13:24:29Z

Dimos at 12:03, 16 October 2012

2012-10-16T12:03:19Z

Dimos: Created page with "Datasets -> Datasets List -> Current Page {| style="width: 100%" |- | align="right" | {| |- | '''Created: '''2012-10-06 |- | {{Last updated}} |} |} =Contact Author=…"

2012-10-06T18:04:23Z

Created page with "Datasets -> Datasets List -> Current Page {| style="width: 100%" |- | align="right" | {| |- | '''Created: '''2012-10-06 |- | {{Last updated}} |} |} =Contact Author=…"

New page

[[Datasets]] -> [[Datasets List]] -> Current Page

{| style="width: 100%"
|-
| align="right" |

{|
|-
| '''Created: '''2012-10-06
|-
| {{Last updated}}
|}

|}

=Contact Author=
Kai Wang
EBU3B, Room 4148
Department of Comp. Sci. and Engr.
University of California, San Diego
9500 Gilman Drive, Mail Code 0404
La Jolla, CA 92093-0404
Email: k[http://mailhide.recaptcha.net/d?k=01omEyRZid0nVm1TN9t98J1A==&c=cUMGBbQUpaP6Zu0AxhNJ8zylFKIVmLYKdQ3GCxUgtxY= ...]@cs.ucsd.edu



----
This page is editable only by [[IAPR-TC11:Reading_Systems#TC11_Officers|TC11 Officers ]].

@@ Line 24: / Line 24: @@
 =Current Version=
 .0 (also available from the [http://vision.ucsd.edu/~kai/svt/ Author's Web site])
 =Keywords=
 OCR, Real Scene, Urban Scene, Scene Text, Word Spotting, Scene Text Recognition, Scene Text Detection, Scene Text Localization

@@ Line 38: / Line 38: @@
 * full image lexicon-driven word detection and recognition.
-If you need character training data then you should look into the Chars74K and ICDAR datasets.
+If you need character training data then you should look into the [http://vision.ucsd.edu/~kai/svt/#related Chars74K] and the [http://www.iapr-tc11.org/mediawiki/index.php/ICDAR_2003_Robust_Reading_Competitions ICDAR2003] and [http://www.iapr-tc11.org/mediawiki/index.php/ICDAR_2005_Robust_Reading_Competitions ICDAR2005] datasets.
-<!--
+=Metadata and Ground Truth Data=
-=Metadata and Ground Truth Data=
+==Harvest images==
-TODO
+Workers are assigned a unique city and are requested to acquire 20 images that contain text from Google Street view. They were instructed to: (1) perform a Search Nearby:* on their city, (2) examine the businesses in the search results, and (3) look at the associated street view for images containing text from the business name. If words are found, they compose the scene to minimize skew, save a screen shot, and record the business name and address.
-=Related Tasks=
+The annotations are in XML using tags similar to those from the [http://algoval.essex.ac.uk/icdar/Datasets.html ICDAR 2003 Robust Reading Competition].
 ----
 =References=
-# To DO. [http://www.iapr-tc11.org/dataset/NEOCR/cbdar_paper.pdf (PDF)]
+# Kai Wang, Boris Babenko and Serge Belongie,  "End-to-end Scene Text Recognition", [http://www.iccv2011.org/ ICCV 2011], Barcelona, Spain [http://www.iapr-tc11.org/dataset/SVT/wang_iccv2011.pdf (PDF)]. Galleries: [http://vision.ucsd.edu/~kai/grocr/icdar/gallery.html ICDAR], [http://vision.ucsd.edu/~kai/grocr/svt/gallery.html SVT].
 =Download=
 ==Version 1.0==
-TODO
+* [http://www.iapr-tc11.org/dataset/SVT/svt.zip The complete Street View Text dataset along with annotations] (118 MB)
 ----
 This page is editable only by [[IAPR-TC11:Reading_Systems#TC11_Officers|TC11 Officers ]].

@@ Line 22: / Line 22: @@
   La Jolla, CA 92093-0404
   Email: k[http://mailhide.recaptcha.net/d?k=01omEyRZid0nVm1TN9t98J1A==&c=cUMGBbQUpaP6Zu0AxhNJ8zylFKIVmLYKdQ3GCxUgtxY= ...]@cs.ucsd.edu
 =Current Version=
-[[Image:neocr_examples.jpg|400px|thumb|right| Example images from the NEOCR dataset. Note that the dataset also includes images with text in different languages, text with vertical character arrangement, light text on dark and dark text on light background, occlusion,
+[[Image:StreetViewText_Sample.jpg|400px|thumb|right| Example images from the Street View Text dataset.]]
-.0 (also available from the [http://www6.cs.fau.de/research/projects/pixtract/neocr/ NEOCR Web site])
+.0 (also available from the [http://vision.ucsd.edu/~kai/svt/ Author's Web site])
 =Keywords=
-OCR, Natural Scene, Scene Text, Word Spotting, Scene Text Recognition, Scene Text Detection, Scene Text Localization
+OCR, Real Scene, Urban Scene, Scene Text, Word Spotting, Scene Text Recognition, Scene Text Detection, Scene Text Localization
 =Description=
-The NEOCR dataset contains 659 real world images with 5238 annotated bounding boxes (textfields). The images were taken by several people independently from the dataset, so the dataset covers a broad range of characteristics which distinguish real world images from scanned documents. All text recognizable by humans has been annotated for all images. The dataset creation process was stopped when for each metadata dimension at least 100 textfields were included in the dataset.
+The Street View Text (SVT) dataset was harvested from Google Street View. Image text in this data exhibits high variability and often has low resolution. In dealing with outdoor street level imagery, we note two characteristics. (1) Image text often comes from business signage and (2) business names are easily available through geographic business searches. These factors make the SVT set uniquely suited for word spotting in the wild: given a street view image, the goal is to identify words from nearby businesses. More details about the data set can be found in our paper, Word Spotting in the Wild [[#References|[1]]]. For our up-to-date benchmarks on this data, see our paper, End-to-end Scene Text Recognition [[#References|[2]]].
-The ground truth contains not only the visible text, but also distortion quadrangles, which enclose the visible text much more precisely than bounding boxes. The dataset is enriched with metadata consisting of brightness, contrast, inversion, texture, resolution, noise, blur, distortion, rotation, character arrangement, occlusion, typeface and language information. The annotation is provided in XML based on the schema of LabelMe.
+This dataset only has word-level annotations (no character bounding boxes) and should be used for
-=Metadata and Ground Truth Data=
+If you need character training data then you should look into the Chars74K and ICDAR datasets.
-Global image metadata includes the filename, folder, source information, image width, height, depth, brightness and contrast. Textfield (local, bounding box) metadata contains the visible text and optical, geometrical and typographical characteristics. Bounding boxes are rectangular and parallel to the axes. Additionally distortion quadrangles are provided which enclose the visible text more precisely.
+<!--
-Optical characteristics include texture, brightness, contrast, inversion, resolution, noise and blur information. Texture, noise and inversion were annotated manually, the rest was computed automatically using ImageMagick. Geometrical characteristics cover distortion, rotation, character arrangement and occlusion information. Typographical characteristics contain typeface and language metadata. Please see the CBDAR paper [[#References|[1]]], the technical report [[#References|[2]]] or the [http://www.iapr-tc11.org/dataset/NEOCR/neocr_metadata_doc.pdf metadata documentation] for further details on the metadata.
+=Metadata and Ground Truth Data=
 =Related Tasks=
@@ Line 54: / Line 51: @@
 =References=
-# R. Nagy, A. Dicker and K. Meyer‐Wegener, "NEOCR: A Configurable Dataset for Natural Image Text Recognition". In CBDAR Workshop 2011 at ICDAR 2011. pp. 53‐58, September 2011. [http://www.iapr-tc11.org/dataset/NEOCR/cbdar_paper.pdf (PDF)], [http://www.iapr-tc11.org/dataset/NEOCR/cbdar_presentation.pdf (Presentation)]
+# To DO. [http://www.iapr-tc11.org/dataset/NEOCR/cbdar_paper.pdf (PDF)]
-# R. Nagy, A. Dicker, and K. Meyer‐Wegener, "Definition and Evaluation of the NEOCR Dataset for Natural‐Image Text Recognition". University of Erlangen, Dept. of Computer Science, Technical Reports, CS‐2011‐07, September 2011. [http://www.iapr-tc11.org/dataset/NEOCR/neocr_techrep.pdf (PDF)]
+=Download=
 ==Version 1.0==
 * [http://www.iapr-tc11.org/dataset/NEOCR/neocr_dataset.tar.gz The complete NEOCR dataset with annotations] (1.3 GB)
 -->