Health care on food and good practices in dietary behavior are drawing people’s attention recently.
Nowadays technology can support the users in keep tracks of their food consumption, and to  increase the awareness in their daily diet by monitoring their food habits. In the recent years many research works have demonstrated that computer vision techniques can help to automatically recognize diverse foods and to estimate the food quantity. Both these two goals are fundamental for a comprehensive diet monitoring system.

We have designed datasets and algorithms for automatic dietary monitoring of canteen customers based on robust computer vision techniques.

UNIMIB2015 Food Database

This database can be used for food recognition and leftoevr estimation. Used in our paper “” where we built a complete system for food logging in a canteen environment. The database is composed of 2,000 tray images with multiple foods and containing 15 food categories. The images are paired with the corresponding empy trays that can be used for leftover estimation.

UNIMIB2016 Food Database

This database can be used for food recognition. The database is composed of 1,027 tray images with multiple foods and containing 73 food categories.

Download UNIMIB2016

• To download the images follow this link:

    UNIMIB2016-images.zip (2.4 GiB, 15,348 hits)

• To download the annotations follow this link:

    UNIMIB2016-annotations.zip (953.1 KiB, 3,213 hits)

• To download the training/test splits follow this link:

    split.zip (48.7 KiB, 1,965 hits)

If you use this database, please cite the following paper:

@article{cioccaJBHI,
 author = {Ciocca, Gianluigi and Napoletano, Paolo and Schettini, Raimondo},
 year = {2017},
 pages = {588-598},
 volume = {21},
 number = {3},
 title = {Food recognition: a new dataset, experiments and results},
 publisher = {IEEE},
 journal = {IEEE Journal of Biomedical and Health Informatics},
 doi = {10.1109/JBHI.2016.2636441}}

@article{cioccaJBHI,
 author = {Ciocca, Gianluigi and Napoletano, Paolo and Schettini, Raimondo},
 year = {2017},
 pages = {588-598},
 title = {Food recognition: a new dataset, experiments and results},
 volume = {21},
 number = {3},
 publisher = {IEEE},
 journal = {IEEE Journal of Biomedical and Health Informatics},
 pdf = {/download/JBHI2016_r1.pdf},
 doi = {10.1109/JBHI.2016.2636441},
 projectref = {http://www.ivl.disco.unimib.it/activities/food-recognition/}}

@inproceedings{ciocca2015food-recognition,
 author = {Ciocca, Gianluigi and Napoletano, Paolo and Schettini, Raimondo},
 editor = {Murino, Vittorio and Puppo, Enrico and Sona, Diego and Cristani, Marco and Sansone, Carlo},
 year = {2015},
 pages = {334-341},
 title = {Food Recognition and Leftover Estimation for Daily Diet Monitoring},
 volume = {9281},
 publisher = {Springer International Publishing},
 series = {Lecture Notes in Computer Science},
 booktitle = {New Trends in Image Analysis and Processing -- ICIAP 2015 Workshops},
 doi = {10.1007/978-3-319-23222-5_41}}