   Github link https://github.com/mhelhoseiny/sherlock 1) 6DS benchmark (28,000 images, 186 unique facts) wit the training and testing splits Fact Recognition Top 1 Accuracy (our method): 69.63% Fact Recognition MAP/MAP100 (our method): 34.86%/ 50.68% 2)  LSC (Large Scale benchmark)  (814K images, 202K unique  facts) with the training and testing splits part 1,LSC_dataset.tar.gz.aa  part 2, LSC_dataset.tar.gz.ab After download cat LSC_dataset.tar.gz.* > LSC_dataset.tar.gz Then extract LSC_dataset.tar.gz Models 1) Model 2 trained on LSC benchmark  caffemodel  deploy_prototxt Feel free to contact for any questions. References  [1] Mohamed Elhoseiny, Scott Cohen, Walter Chang, Brian Price, Ahmed Elgammal, Sherlock: Scalable Fact Learning in Images, AAAI, 2017, acceptance rate (24%).  [2] Mohamed Elhoseiny, Scott Cohen, Walter Chang, Brian Price, Ahmed Elgammal, Automatic Annotation of Structured Facts in Images, ACL Proceedings of the  Vision&Language Workshop, 2016 (long paper) |