- Rating
- 1
- Confidence
- 3
- Informativeness
- 2
- Presentation
- 2
- Soundness
- 2
Read review
Paper summary
The paper studies budget-constrained learned image compression through a unified lightweight variable-rate architecture. It first adjusts compute allocation between transforms and entropy modeling, then searches modules locally and composes them, and finally builds a model family through expansion and energy-based hierarchical pruning. Kodak and CLIC-PV results show favorable rate–distortion versus MAC trade-offs across several budgets. Experiments also report NVIDIA L20 timings and compare pruning with extended from-scratch training.
Critical strengths
-
The study goes beyond a single compact codec and reports variable-rate models across a broad range of decoder budgets. For example, M1× improves compression substantially over Shallow-NTC while reducing encoder and decoder MACs. This makes the model family a useful empirical contribution to low-complexity compression.
-
Comparisons with single-step pruning and extended from-scratch training help distinguish parameter-transfer benefits from simply spending more optimization effort. The observation that twice-longer from-scratch training approaches hierarchical-pruning performance is informative and appropriately narrows the interpretation to convergence efficiency.
Critical weaknesses
-
The entropy/transform conclusion is derived mainly from 1/2/4-step variants near one 20-kMAC design, with transform width changing as entropy capacity changes. Local search then restricts each module to a narrow complexity range inherited from the base architecture, limiting opportunities to discover different allocations. Scaling and pruning that architecture do not by themselves show that the same procedure finds good allocations across budgets. Without matched search controls, the evidence supports a successful architecture recipe more strongly than a general method for deciding where compute should go.
-
The motivation is constrained edge hardware, but timings are measured on a server GPU. Table 1 illustrates the limitation of the proxy: M1× has fewer decoder MACs than Shallow-NTC but takes 9.7 rather than 6.3 ms. Sequential entropy passes and operator efficiency can change the ranking. Consequently, favorable rate–distortion/MAC curves are insufficient to substantiate the targeted smartphone practicality, a central part of the paper's framing.
Questions for authors
-
Does allocation plus local search outperform joint or random search at equal total training compute across multiple decoder budgets? How sensitive are the selected allocations to the initial architecture?
-
Do the reported gains persist for complete codec latency on the intended smartphone or a constrained CPU, including entropy coding and data movement? How should the latency reversal against Shallow-NTC be interpreted?
2 comments
Response to Reviewer AxpP (2/2): Complete smartphone evaluation and decoding performance
Thank you for recognizing the compression gains across our decoder budgets. This second response addresses complete smartphone execution (Q2/W2); Part 1 covers model-family design and cross-budget controls.
We have revised the manuscript, with changes in blue. Figure 4 and Section 4.3 now present the family's measured resource and smartphone decoding performance.
Questions for Authors
Q2. Complete codec latency, entropy processing and data movement
The family's advantages extend to complete smartphone decoding, including CPU entropy processing and required tensor movement. x1/x2/x4/x8 improve native RD with fewer decoder MACs and parameters than Shallow-NTC/EVC-S/EVC-M/EVC-L, respectively, and decode faster on both tested phone images. x0.25 also improves RD and full codec latency over Shallow-NTC.
Paired values are Kodak/CLIC-PV (K/C). MAC/Param denotes decoder kMAC/pixel/M parameters. BD uses native actual-stream dataset RD relative to x1 over shared quality intervals; lower is better. Phone API times are medians.
| Model | MAC/Param | BD %, K/C | Full encode ms, K/C | Full decode ms, K/C |
|---|---|---|---|---|
| Ours x0.25 | 4.94/1.10 | +15.96/+18.84 | 40.5/256.8 | 34.8/227.7 |
| Ours x1 | 17.94/2.51 | 0/0 | 45.8/363.3 | 42.9/286.5 |
| Shallow-NTC | 20.51/10.47 | +19.93/+33.11 | 205.3/1627.8 | 78.1/561.8 |
| Ours x2 | 43.51/2.99 | −6.37/−7.55 | 59.8/493.7 | 50.0/354.1 |
| EVC-S | 69.56/10.92 | −1.22/+0.34 | 84.3/709.7 | 72.5/626.8 |
| Ours x4 | 81.34/6.25 | −9.54/−11.09 | 90.6/754.8 | 66.4/501.9 |
| EVC-M | 157.56/12.11 | −6.78/−5.67 | 117.2/1395.1 | 81.4/693.3 |
| Ours x8 | 160.79/12.19 | −13.07/−14.97 | 127.8/1056.6 | 75.4/620.0 |
| EVC-L | 291.31/13.16 | −8.56/−7.61 | 172.6/2203.5 | 109.6/868.9 |
Tests use iPhone 17 Pro Max/CoreML ALL on kodim01 (512×768) and one CLIC image (1363×2048), H×W, at the model-specific operating points in Table A11. Complete APIs include CPU entropy coding/decoding, CDF/index/quantization, neural execution, required tensor movement and CPU-readable RGB8 output. Loading, compilation and disk I/O are excluded.
Appendix E.3 reports all configurations (Tables A12–A13), memory and explicit I/O (Table A14), and measured mean component shares (Figure 19). CoreML-internal movement remains included within CoreML/total time. Dataset RD and the two-image phone timings retain their respective scopes.
Responses to Weaknesses
W2. Practical deployment and the latency reversal
The original L20 neural times remain 9.7/6.3 ms for x1/Shallow-NTC, while x0.25 takes 3.5 ms. Sequential context execution and backend/operator efficiency affect latency beyond MACs. The L20 neural timings and complete phone APIs use different hardware, backends and timing scopes; these measurements do not isolate a single cause of the ranking change.
The new evidence directly tests the intended device and includes the entropy processing and data movement requested in Q2. Figure 4 jointly evaluates RD, parameters, computation, neural decoding and complete decoding, demonstrating the family's practical choices across budgets.
Response to Reviewer AxpP (1/2): Model-family design and cross-budget controls
Thank you for recognizing our model family's compression advantages across decoder budgets.
We have revised the manuscript, with changes in blue. Our contribution is a unified framework for constructing structurally compatible codec families across budgets. This response addresses allocation, search and initialization (Q1/W1); Part 2 presents complete smartphone measurements and the family's decoding benefits (Q2/W2).
Questions for Authors
Q1. Multiple budgets, training compute and initialization
Local search and hierarchical transfer serve model-family construction. The controls below assess allocation choices, search strategy and dependence on the starting architecture.
Allocation across budgets. Within each budget, encoder architecture is fixed and weights are independently trained. The added controls give the following BD-rates relative to the preselected 2-step references (Kodak/CLIC; Appendix E.2, Tables A6–A9):
The preferred tested allocation changes with budget. At 20k, 4-step has higher BD-rate than 2-step in all three seeds (mean +5.680/+5.759% relative to 2-step).
Search controls. Local and Random use the same candidate space and 20k-update candidate training, approximately matched search-process GPU time, and 1M final-training updates. Total training GPU time and actual MACs are not exactly matched. Random samples joint configurations; no other joint-search optimizer is tested.
The table summarizes three paired runs per budget. BD is Local relative to Random; negative favors Local. Cost change uses cumulative search-plus-final-training GPU time within each budget.
The results do not establish a consistent RD advantage for Local. Neither strategy wins consistently at 8k/20k; Random wins all three 40k pairs. At 8k, paired Kodak BD ranges from −13.262% to +15.134%. Tables A1–A4 report all nine paired results, actual MACs and available cost breakdowns. The 8k/20k runs use A100; 40k uses H20.
Initialization. Changing the starting module configuration can improve or worsen final RD. Five completed runs from two alternative starting configurations, relative to original-start Local at the same seed, span −1.164% to +1.411% on Kodak and −1.690% to +1.271% on CLIC (Table A5).
Responses to Weaknesses
W1. Allocation scope and local candidate ranges
Sections 3.2–3.4 connect budget-guided allocation and module refinement to expansion that preserves block/channel correspondences for parameter transfer. The construction specifies both budget-specific architectures and their structural relationships.
The new allocation controls extend the analysis beyond the original 20k design. Local retains the base architecture's candidate ranges, and module composition need not improve monotonically. The search controls characterize this construction step within the tested settings.
The resulting family's RD and complete-decoding benefits are evaluated directly in revised Figure 4 and Part 2.