EgocentricFood contains images taken by a wearable egocentric vision camera. The pictures focus on scenes where some kind of food or food-related object appears, including a total of 9 different food-related classes, each of which has a bounding box indicating the location of the food item in the photo.
This dataset can be used for two different tasks: food recognition (only using the image crops indicated by the bounding boxes), or food detection/localization.
Contents
This ZIP file contains the following files:
- [1-9]: directory names correspond to food-related object IDs.
- [1-9]/*.jpg: food photo files (some photos are duplicated in two or more directories, since they include two or more food items.)
- bb/bb_NARRATIVE_[1-9].txt: bounding box information for the photo files in each directory
- category.txt: food list including the correspondences between food IDs and
food names in English - [train/val/test]_list.txt: list of images belonging to the [train/val/test] split, including image path, category ID and image coordinates
Citation
If you use this dataset for any purpose, please, do not forget to cite the following paper:
Bolaños M. and Radeva P. «Simultaneous Food Localization and Recognition» In Proceedings of the 23rd International Conference on Pattern Recognition (ICPR) 2016 (IN PRESS).
All the pictures contained in the dataset have been anonymized by blurring all the faces after the publication of the cited paper. This procedure has been applied in order to avoid any possible ethical issues. For this reason, we can not guarantee that the results presented in the paper can be precisely reproduced.
If you have any doubt or proposal, please, do not hesitate to contact the first author:
Contact
Marc Bolaños
marc.bolanos@ub.edu