The Image Calculator: 10x Faster Image-AI Inference by Replacing JPEG with Self-designing Storage Format
Summary: Image Calculator: a self-designing image storage format that adapts to the AI task by exploring a large space of candidate formats with composite performance models. Outperforms JPEG variants with up to 14.2x faster inference and 8.2x storage reduction, with minimal accuracy loss, across diverse data, models, and hardware. (summarized by gpt-5-nano on Feb 09 2026)
Incoming Non-self Citations Over Time
Authors
- 1. Utku Sirin (Harvard University)
- 2. Stratos Idreos (Harvard University)
BibTeX Citation
@inproceedings{sirin_sigmod24,
title = {{The Image Calculator: 10x Faster Image-AI Inference by Replacing JPEG with Self-designing Storage Format}},
author = {Sirin, Utku and Idreos, Stratos},
series = {{SIGMOD} '24},
booktitle = {Proceedings of the {ACM} {SIGMOD} International Conference on Management of Data},
publisher = {Association for Computing Machinery},
doi = {10.1145/3639307},
url = {https://dl.acm.org/doi/10.1145/3639307},
year = {2024}
}Incoming Citations (Sorted by Pagerank)
Showing 3 of 3 citing papers.
| Rank | Citing Paper | Year | Venue | Pagerank |
|---|---|---|---|---|
| 6,166 | Apt-Serve: Adaptive Request Scheduling on Hybrid Cache for Scalable LLM Inference Serving | 2025 | SIGMOD | 5.8631131e-05 |
| 8,620 | Limousine: Blending Learned and Classical Indexes to Self-Design Larger-than-Memory Cloud Storage Engines | 2024 | SIGMOD | 5.301557e-05 |
| 9,987 | Frequency-Store: Scaling Image AI by A Column-Store for Images | 2025 | CIDR | 5.1001827e-05 |
Previous
Page 1 / 1
Next
Outgoing Citations (Sorted by Pagerank)
Showing 50 of 72 cited papers.
Citations counted here include only citations to other VLDB/SIGMOD/CIDR/PODS papers in this database.
Semantically Similar Papers
| # | Overall Rank | Paper | Year | Venue |
|---|---|---|---|---|
| 1 | 1,614 | Compressed Linear Algebra for Large-Scale Machine Learning | 2016 | VLDB |
| 2 | 3,140 | DeepSqueeze: Deep Semantic Compression for Tabular Data | 2020 | SIGMOD |
| 3 | 11,307 | Inference-friendly Graph Compression for Graph Neural Networks | 2025 | VLDB |
| 4 | 3,040 | Incremental and Approximate Inference for Faster Occlusion-based Deep CNN Explanations | 2019 | SIGMOD |
| 5 | 9,071 | Privacy and Accuracy-Aware AI/ML Model Deduplication | 2025 | SIGMOD |
| 6 | 13,664 | Alphabets, Grammars, Calculators, and the End of Hand-Crafted Systems | 2025 | VLDB |
| 7 | 10,146 | QStore: Quantization-Aware Compressed Model Storage | 2026 | VLDB |
| 8 | 3,219 | Jointly Optimizing Preprocessing and Inference for DNN-based Visual Analytics | 2021 | VLDB |
| 9 | 6,311 | Progressive Compressed Records: Taking a Byte out of Deep Learning Data | 2021 | VLDB |
| 10 | 9,987 | Frequency-Store: Scaling Image AI by A Column-Store for Images | 2025 | CIDR |