The Food Recognition Benchmark: Using DeepLearning to Recognize Food on Images
Authors:
Sharada Prasanna Mohanty,
Gaurav Singhal,
Eric Antoine Scuccimarra,
Djilani Kebaili,
Harris Héritier,
Victor Boulanger,
Marcel Salathé
Abstract:
The automatic recognition of food on images has numerous interesting applications, including nutritional tracking in medical cohorts. The problem has received significant research attention, but an ongoing public benchmark to develop open and reproducible algorithms has been missing. Here, we report on the setup of such a benchmark using publicly available food images sourced through the mobile My…
▽ More
The automatic recognition of food on images has numerous interesting applications, including nutritional tracking in medical cohorts. The problem has received significant research attention, but an ongoing public benchmark to develop open and reproducible algorithms has been missing. Here, we report on the setup of such a benchmark using publicly available food images sourced through the mobile MyFoodRepo app. Through four rounds, the benchmark released the MyFoodRepo-273 dataset constituting 24,119 images and a total of 39,325 segmented polygons categorized in 273 different classes. Models were evaluated on private tests sets from the same platform with 5,000 images and 7,865 annotations in the final round. Top-performing models on the 273 food categories reached a mean average precision of 0.568 (round 4) and a mean average recall of 0.885 (round 3). We present experimental validation of round 4 results, and discuss implications of the benchmark setup designed to increase the size and diversity of the dataset for future rounds.
△ Less
Submitted 30 June, 2021; v1 submitted 28 June, 2021;
originally announced June 2021.
Generator Contingency Modeling in Electric Energy Markets: Derivation of Prices via Duality Theory
Authors:
N. G. Singhal,
J. Kwon,
K. W. Hedman
Abstract:
Traditional electric energy markets do not explicitly model generator contingencies. To improve the representation of resources and to enhance the modeling of uncertainty, existing markets are moving in the direction of including generator contingencies and remedial action schemes within market auction models explicitly. This research contributes to the market design realm by providing detailed an…
▽ More
Traditional electric energy markets do not explicitly model generator contingencies. To improve the representation of resources and to enhance the modeling of uncertainty, existing markets are moving in the direction of including generator contingencies and remedial action schemes within market auction models explicitly. This research contributes to the market design realm by providing detailed analysis of impending changes, it provides insightful guidance in understanding the market implications, and it provides recommendations on necessary changes to ensure a fair and transparent market structure. A primal (and the corresponding dual) formulation that accounts for the proposed changes to the auction model is provided to enable a theoretical analysis of the anticipated changes including the effect on market prices, settlements, and revenues. The derivation of the prices and the dual formulation are based on leveraging duality theory from linear optimization theory. A comparison to existing market structures is also included. The primary impact of the proposed changes includes the addition of a new congestion component within the traditional locational marginal price, which reflects the influence of congestion during the post-contingency states for the modeled critical generator contingencies.
△ Less
Submitted 5 October, 2019;
originally announced October 2019.
