Conditional Positional Encodings for Vision Transformers
This work addresses a key limitation in vision Transformers for computer vision researchers and practitioners, offering an incremental improvement over existing positional encoding methods.
The paper tackles the problem of positional encodings in vision Transformers by proposing a conditional positional encoding (CPE) scheme that dynamically generates encodings based on local input tokens, enabling generalization to longer sequences and improving translation-invariance, resulting in outperforming performance in image classification tasks.
We propose a conditional positional encoding (CPE) scheme for vision Transformers. Unlike previous fixed or learnable positional encodings, which are pre-defined and independent of input tokens, CPE is dynamically generated and conditioned on the local neighborhood of the input tokens. As a result, CPE can easily generalize to the input sequences that are longer than what the model has ever seen during training. Besides, CPE can keep the desired translation-invariance in the image classification task, resulting in improved performance. We implement CPE with a simple Position Encoding Generator (PEG) to get seamlessly incorporated into the current Transformer framework. Built on PEG, we present Conditional Position encoding Vision Transformer (CPVT). We demonstrate that CPVT has visually similar attention maps compared to those with learned positional encodings and delivers outperforming results. Our code is available at https://github.com/Meituan-AutoML/CPVT .