arXiv:2109.07301
What Vision-Language Models `See' when they See Scenes
원문 초록
Images can be described in terms of the objects they contain, or in terms of the types of scene or place that they instantiate. In this paper we address to what extent pretrained Vision and Language models can learn to align descriptions of both types with images. We compare 3 state-of-the-art models, VisualBERT, LXMERT and CLIP. We find that (i) V&L models are susceptible to stylistic biases acquired during pretraining; (ii) only CLIP performs consistently well on both object- and scene-level descriptions. A follow-up ablation study shows that CLIP uses object-level information in the visual modality to align with scene-level textual descriptions.
함께 살펴볼 논문
공통 태그를 근거로 연결했습니다. 실제 인용 관계나 논문의 우수성을 뜻하지 않습니다.
RoHOI: Robustness Benchmark for Human-Object Interaction Detection
Car Object Counting and Position Estimation via Extension of the CLIP-EBC Framework
NOCTIS: Novel Object Cyclic Threshold based Instance Segmentation
Refine Any Object in Any Scene
DenseWorld-1M: Towards Detailed Dense Grounded Caption in the Real World
Towards Reliable Detection of Empty Space: Conditional Marked Point Processes for Object Detection
토론
이 논문과 연결된 포럼 글이 아직 없습니다.