Extending Flexibility of Image Coding Enhancement Framework for IoTs
Keywords: Data Compression, IoT infrastructure, Edge Computing, Scalable Design
Abstract: Neural image compression, necessary in various edge-device scenarios, suffers from its heavy encode-decode structures and inflexible compression level switch. The primary issue is that the computational and storage capabilities of edge devices are weaker than those of servers, preventing them from handling the same amount of computation and storage. One solution is to downsample images and reconstruct them on the receiver side; however, current methods uniformly downsample the image and limit flexibility in compression levels. We take a step to break up this paradigm by proposing a conditional uniform-based sampler that allows for flexible image size reduction and reconstruction. Building on this, we introduce a lightweight transformer-based reconstruction structure to further reduce the reconstruction load on the receiver side. Extensive evaluations conducted on a real-world testbed demonstrate multiple advantages of our system over existing compression techniques, especially in terms of adaptability to different compression levels, computational efficiency, and image reconstruction quality.
Primary Area: infrastructure, software libraries, hardware, systems, etc.
Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.
Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.
Reciprocal Reviewing: I understand the reciprocal reviewing requirement as described on https://iclr.cc/Conferences/2025/CallForPapers. If none of the authors are registered as a reviewer, it may result in a desk rejection at the discretion of the program chairs. To request an exception, please complete this form at https://forms.gle/Huojr6VjkFxiQsUp6.
Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.
No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.
Submission Number: 14271
Loading