A benchmark dataset of food photos with food records for evaluation of computer vision algorithms in the context of dietary assessment

Authors: CHIN, ELIZABETH - University Of California, Davis; Larke, Jules; BOUZID, YASMINE - University Of California, Davis; Nguyen, Tu; VAINBERG, YAEL - University Of California, Davis; SMILOWITZ, JENNIFER - University Of California, Davis; Lemay, Danielle

Submitted to: Ag Data Commons
Publication Type: Database / Dataset
Publication Acceptance Date: 12/9/2022
Publication Date: 12/13/2022
Citation: Chin, E.L., Larke, J.A., Bouzid, Y.Y., Nguyen, T.T., Vainberg, Y., Smilowitz, J.T., Lemay, D.G. 2022. A benchmark dataset of food photos with food records for evaluation of computer vision algorithms in the context of dietary assessment. Ag Data Commons. https://doi.org/10.15482/USDA.ADC/1528346.
DOI: https://doi.org/10.15482/USDA.ADC/1528346

Interpretive Summary: Photo-based dietary assessment methods are becoming more feasible as artificial intelligence methods improve. However, advancement of these methods to the level usable in nutrition studies has been hindered by the lack of a dataset against which to benchmark algorithm performance. We conducted the Surveying Nutrient Assessment with Photographs of Meals (SNAPMe) Study (ClinicalTrials ID: NCT05008653) to develop a benchmark dataset of food photographs paired with traditional food records. The SNAPMe database contains 3,311 unique food photos linked with 275 food records from 95 participants who photographed all foods consumed and recorded food records in parallel for up to 3 study days each. A description of the SNAPMe database as well as its utility as a benchmark will be described in a companion manuscript.

Technical Abstract: Photo-based dietary assessment methods are becoming more feasible as artificial intelligence methods improve. However, advancement of these methods to the level usable in nutrition studies has been hindered by the lack of a dataset against which to benchmark algorithm performance. We conducted the Surveying Nutrient Assessment with Photographs of Meals (SNAPMe) Study (ClinicalTrials ID: NCT05008653) to develop a benchmark dataset of food photographs paired with traditional food records. Participants were recruited nationally and completed enrollment meetings via web-based video conferencing. By the end of the study, 90 participants had completed all three days of data collection; 95 participants completed at least one study day. Participants uploaded and annotated their meal photos using a mobile phone app called Bitesnap and completed food records using the Automated Self-Administered 24-hour Dietary Assessment Tool (ASA24®) on the same day. A sizing marker with black and white boxes of known size were included in meal photos. Participants included photos “before” and “after” eating non-packaged and multi-serving packaged meals, as well as photos of the “front” package label and “ingredient” label for single-serving packaged foods. In total, the SNAPMe Database (DB) contains 3,311 unique food photos linked with 275 ASA24 food records from 95 participants who photographed all foods consumed and recorded food records in parallel for up to 3 study days each for a total of 275 diet days. The SNAPMe DB includes 1,475 “before” photos of non-packaged foods, 1,436 “after” photos of non-packaged foods, 203 “front” photos of packaged foods, and 196 “ingredient” labels of packaged foods. Each line item of each ASA24 food record is linked to the relevant photo. These data will be transformative for the improvement of artificial intelligence algorithms for the adoption of photo-based dietary assessment in nutrition research