AI Paper Atlas

Research map / Vision-language and multimodal / Human and embodied modeling

Large Scale Visual Food Recognition

Weiqing Min, Zhiling Wang, Yuxin Liu, Mengjiang Luo, Liping Kang, Xiaoming Wei, Xiaolin K. Wei, Shuqiang Jiang

TPAMI 2023 · Vision-language and multimodal › Human and embodied modeling · approach: Scene graph generation · 296 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
increase learning stages U from 1 to 3 (ablation)PRENet(ResNet50) · Food2Ktop-1 accuracy (%): 81.45 -> 82.11 -> 83.03Improved
increase learning stages to U=4 (ablation)PRENet(ResNet50) · Food2Ktop-1 accuracy (%)Hurt
add progressive learning strategy (PL) to simple feature fusion (ablation)PRENet(ResNet50) · Food2Ktop-1 accuracy (%)Improved
add region feature enhancement (RE) on top of PL (ablation)PRENet(ResNet50) · Food2Ktop-1 accuracy (%)Improved
use only cross-entropy loss (alpha=1) instead of adding KL divergence (ablation)PRENet(ResNet50) · Food2Ktop-1 accuracy (%): reduced by 3.18%Improved
use only KL divergence (alpha=0, beta=1) (ablation)PRENet(ResNet50) · Food2Ktop-1 accuracy (%)Hurt
add PRENet to ResNet50 backboneResNet50 · Food2Ktop-1 accuracy (%): 80.79 -> 83.03Improved
pre-fine-tune on Food2K before target datasetVGG16 · Vireo Food-172top-1 accuracy (%): 80.41 -> 87.74Improved
pre-fine-tune on Food2K before target datasetVGG16 · ETH Food-101top-1 accuracy (%): 79.02 -> 80.68Improved
pre-fine-tune on Food2K instead of ETH Food-101Dynamic R-CNN · UNIMIB2016 detectionmAP (%): 69.2 -> 70.6Improved
fine-tune on ETH Food-101 for metric-learning retrievalTriplet loss ResNet101 · Vireo Food-172 retrievalmAP (%): 83.14 -> 77.68Hurt
fine-tune backbone on Food2K for PSPnetPSPnet(ResNet50) · UEC-FoodPix CompletemIoU (%): 70.9 -> 74.5Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy