• Publications
  • Influence
Women also Snowboard: Overcoming Bias in Captioning Models
TLDR
We introduce a new Equalizer model that ensures equal gender probability when gender evidence is occluded in a scene and confident predictions whengender evidence is present. Expand
  • 116
  • 8
  • PDF
Object Hallucination in Image Captioning
TLDR
We propose a new image relevance metric to evaluate current captioning models with veridical visual labels and assess their rate of object hallucination. Expand
  • 43
  • 5
  • PDF
SPLAT: Semantic Pixel-Level Adaptation Transforms for Detection
TLDR
We propose a Semantic Pixel-Level Adaptation Transform (SPLAT) approach to detector adaptation that efficiently generates cross-domain image pairs and exploits pixel-level semantic labels to condition the training of the transformation network. Expand
  • 8
  • 2
  • PDF
Evaluating Theory of Mind in Question Answering
TLDR
We propose a new dataset for evaluating question answering models with respect to their capacity to reason about beliefs. Expand
  • 11
  • PDF
Exploiting Attention to Reveal Shortcomings in Memory Models
TLDR
We analyze the attention of a memory network model to reconcile contradictory performance on a challenging question-answering dataset that is inspired by theory-of-mind experiments. Expand
  • 3
  • PDF
Addressing and Understanding Shortcomings in Vision and Language
Addressing and Understanding Shortcomings in Vision and Language