List your company

Widget Captioning on Widget Captioning (test)

159.8CIDErPaLI-3
66.7290.885115.05139.215Jun 5, 2021Nov 14, 2021Apr 26, 2022Oct 6, 2022Mar 18, 2023Aug 28, 2023Feb 7, 2024

Evaluation results

MethodLinks
PaLI-3OCR Usage=Without OCR2024.02159.8——————
ScreenAIOCR Usage=Without OCR2024.02156.4——————
Pix2Struct LargePretraining=Screenshot parsing, Pixel only=true2022.10136.7——————
Pix2Struct BasePretraining=Screenshot parsing, Pixel only=true2022.10133.1——————
DonutPretraining=OCR, Pixel only=true2022.10127.4——————
Learnable-Fourier+MLPdimension_split=4/12021.06100.746.833.446.132.417.8—
Fixed-Fourier+MLPdimension_split=2/22021.06100.246.132.545.832.518.4—
Learnable-Fourier+MLPdimension_split=1/42021.0699.145.632.745.232.217.1—
Learnable-Fourier+MLPdimension_split=2/22021.069846.132.745.932.617.9—
Fixed-Fourier+MLPdimension_split=4/12021.0697.245.532.145.131.717.6—
SOTA2021.069744.932.244.731.717.6—
Embed-4Ddescription=reproduction of SOTA [20]2021.069745.231.94531.717.3—
VUTtype=State of the art w/ pipelines2022.1097——————
Fixed-Fourier+MLPdimension_split=1/42021.0695.44532.144.231.217.1—
Learnable-Fourierdimension_split=2/22021.0695.344.931.644.331.617.7—
Sine-4D2021.0694.944.931.943.93116.7—
MLP2021.0670.33423.533.723.710.2—
Ferret-UI-13B2024.09——————142
MM1.5-UI-1B2024.09——————144.4
MM1.5-UI-30B2024.09——————145.9
MM1.5-UI-3B2024.09——————145
MM1.5-UI-3BTraining epochs=1 ep.2024.09——————145.2
MM1.5-UI-3BTraining epochs=1 ep., SFT strategy=w/o MM1.5 SFT2024.09——————139.5
MM1.5-UI-7B2024.09——————149.7
PaliGemma-3BFine-tuning protocol=per-task fine-tuning2024.09——————148.4
Spotlight2024.09——————141.8