AIR automatically generates pre-labeled images with Stable Diffusion and GPT-3.5 prompts, then trains classifiers, claiming 95.48% accuracy on generated forest-fire images and a 4% gain on MIT-67.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Automated Image Recognition Framework
AIR automatically generates pre-labeled images with Stable Diffusion and GPT-3.5 prompts, then trains classifiers, claiming 95.48% accuracy on generated forest-fire images and a 4% gain on MIT-67.