Transformer-based Variable-rate Image Compression with Region-of-interest Control
This work addresses efficient image compression for applications requiring flexible bitrate and focus control, representing an incremental advance by integrating variable-rate and ROI functionalities into a learned system.
The paper tackles image compression by proposing a transformer-based system that achieves variable-rate compression and region-of-interest control with a single model, using prompt generation networks to condition the autoencoder based on input image, ROI mask, and rate parameter, and it shows superiority over competing methods in experiments.
This paper proposes a transformer-based learned image compression system. It is capable of achieving variable-rate compression with a single model while supporting the region-of-interest (ROI) functionality. Inspired by prompt tuning, we introduce prompt generation networks to condition the transformer-based autoencoder of compression. Our prompt generation networks generate content-adaptive tokens according to the input image, an ROI mask, and a rate parameter. The separation of the ROI mask and the rate parameter allows an intuitive way to achieve variable-rate and ROI coding simultaneously. Extensive experiments validate the effectiveness of our proposed method and confirm its superiority over the other competing methods.