id	author	title	date	pages	extension	mime	words	sentences	flesch	summary	cache	txt
work_5vqe5iapgfam3egdt4zhjfgrdi	Wai Lam Hoo	Keybook: Unbias object recognition using keywords	2015	9	.pdf	application/pdf	5996	648	62	The presence of bias in existing object recognition datasets is now a well-known problem in the computer Efros (2011) raised an important question – ''how well does a typical object detector trained on one dataset generalize when tested on critical as it causes one object representation and recognition algorithm will only work well in the specific dataset that one choose to features for an object class that are generalized across all datasets. proposing Keybook generation with the aim to discover the keywords from codebook that learned from the different datasets. In general, keywords are the visual features that significantly represent the respective object classes. The collected Keybook that is generalized over the datasets will greatly enhance the object recognition the domain bias problem will deteriorate the overall object recognition system performance, especially when the trained model is a framework that undo the domain bias thru finding the most significant features of an object class which are generalized across all	./cache/work_5vqe5iapgfam3egdt4zhjfgrdi.pdf	./txt/work_5vqe5iapgfam3egdt4zhjfgrdi.txt