Fast Image Caption Generation with Position Alignment

Understand

Recent neural network models for image captioning usually employ an encoder-decoder architecture, where the decoder adopts a recursive sequence decoding way.

• However, such autoregressive decoding may result in sequential error accumulation and slow generation which limit the applications in practice.
• Non-autoregressive (NA) decoding has been proposed to cover these issues but suffers from language quality problem due to the indirect modeling of the target distribution.
• Towards that end, we propose an improved NA prediction framework to accelerate image captioning.