Details

• The 4 categories of this dataset are:
  
  - Bikes: 365 images Download
  - Cars: 420 images Download
  - Person: 311 images Download
  - None: 380 images Download
  
  
• Ground truth is available for 300 images of each category
  Images with ground truth Download
  The ground truth is given in terms of pixel segmentation masks with values between 0 and 255. Where pixels with 0 denote the object in the image.

---

Related papers

• Generic object recognition with boosting. A. Opelt, M. Fussenegger, A. Pinz and P. Auer. Technical Report TR-EMT-2004-01, EMT, TU Graz, Austria, 2004. Download PDF
• Marcin Marszałek and Cordelia Schmid. Accurate Object Localization with Shape Masks. IEEE Conference on Computer Vision & Pattern Recognition, 2007. Download PDF

---

Evaluation Protocol

Statistics

Class	File	Images			Objects
		total	training	testing
Bikes	ig02-v1.0-bikes.zip (208 MB)	365	162	162	511
Cars	ig02-v1.0-cars.zip (205 MB)	420	177	177	770
People	ig02-v1.0-people.zip (152 MB)	311	140	140	785

---

Features and top scores

Method:	[Paper Name]	[Paper Name]
Score: