Timezone: »

Dynamic Grained Encoder for Vision Transformers
Lin Song · Songyang Zhang · Songtao Liu · Zeming Li · Xuming He · Hongbin Sun · Jian Sun · Nanning Zheng

Wed Dec 08 12:30 AM -- 02:00 AM (PST) @

Transformers, the de-facto standard for language modeling, have been recently applied for vision tasks. This paper introduces sparse queries for vision transformers to exploit the intrinsic spatial redundancy of natural images and save computational costs. Specifically, we propose a Dynamic Grained Encoder for vision transformers, which can adaptively assign a suitable number of queries to each spatial region. Thus it achieves a fine-grained representation in discriminative regions while keeping high efficiency. Besides, the dynamic grained encoder is compatible with most vision transformer frameworks. Without bells and whistles, our encoder allows the state-of-the-art vision transformers to reduce computational complexity by 40%-60% while maintaining comparable performance on image classification. Extensive experiments on object detection and segmentation further demonstrate the generalizability of our approach. Code is available at https://github.com/StevenGrove/vtpack.

Author Information

Lin Song (Xi'an Jiaotong University)
Songyang Zhang
Songtao Liu (Beihang University, Beijing, China)
Zeming Li (Megvii(Face++) Inc)
Xuming He (ShanghaiTech University)
Hongbin Sun (Xi'an Jiaotong University)
Jian Sun (Megvii, Face++)
Nanning Zheng (Xi'an Jiaotong University)

More from the Same Authors