Choose the type closest to your main contribution, then adapt the plan. The seven possible editorial routes distinguish architecture, training/objectives, theory, empirical analysis, resources, efficiency and applications. These are not official ECCV tracks, and a paper can span several routes.
For a new design, start with a readable architecture or representation overview, a comparison table with matched conditions, and controls testing the changed component. Add pseudocode when state, iteration, sampling or gradient routing is otherwise ambiguous; add a theorem only when making a formal claim. Neither is required merely to resemble an oral paper.
Observed medians below describe the inspected source versions. Suggested artifact quantities are editorial starting packages, not measured prescriptions or submission requirements. Small categories are examples, not robust estimates of a conference-wide standard.
The central deliverable changes the model's representation, component arrangement, information flow or reusable inference pipeline. Data, objectives and efficiency can still be substantial supporting contributions.
Corpus context · 10 papers: median 6.5 main figures and 4 main tables; body span ≈ 15.15 pages; attached appendix span ≈ Not resolved pages (0 resolved boundaries). Separate source files: median Not resolved whole-file pages among 0 papers; these can repeat main text and are not appendix-only lengths. Extraction-assisted observations—not drafting targets.
The selected papers define a representation or missing capability before the end-to-end pipeline. Training-only teachers, geometric postprocessing and preserved foundation-model abilities are material parts of the comparison, not caption decoration.
Editorial starting package: choose artifacts that test your actual claims. Quantities below are planning suggestions, not counts inferred from all category members or conference rules.
Start with one overview; add insets only when they resolve different questions
Label inputs, outputs, inherited versus new components, frozen/trainable blocks and training-only paths.
When: Core when the novelty is computational organization; use a simple representation cartoon before the full diagram if the data object is unfamiliar.
One compact example or aligned qualitative grid
Make the claimed representational problem visible and reuse the same example through the mechanism.
When: Especially useful for 3D geometry, multimodal tokens, articulated objects or structured outputs; do not add a gallery that does not test the claim.
One focused table per incompatible task or information regime
Keep metric direction, data, backbone, reference inputs and refinement policy next to the results.
When: Separate single-model inference from per-instance optimization and privileged-input baselines.
A compact switch table or matched diagnostic panels
Test the named reason for each component rather than removing modules without preserving compute or capacity.
When: Use matched controls when the mechanism claim depends on architecture rather than more parameters or more supervision.
Only when the procedure or guarantee requires it
Expose iteration, state, branching or assumptions not recoverable from the diagram.
When: A feed-forward module does not require pseudocode or a theorem simply to look like an accepted paper.
Tensor and layer specifications, training recipes, calibration or geometric recovery details, matched extra results and failure examples. Preserve the defining representation and inference inputs in the main text.
Training supervision and optional inference outputs are explicit; empty-token controls test extra capacity.
Architecture simplicity is kept distinct from distillation and added data.
Boundary: This route contains 10 primary editorial assignments in the 28-paper inspected subset (28 inventory records). It is not an official track, a universal writing standard or evidence that these choices caused acceptance.
See inline examples of these artifact styles · Compare every member in the measurement ledger
The contribution reorganizes a visual-odometry pipeline around a focused plenoptic model and joint photometric/depth residuals.
The principal contribution is a geometric column-to-line yaw-voting pipeline, supported by an invariance proposition and localization tests.
The main deliverable changes where language enters a frozen visual encoder, with paired tests of steerability and representation preservation.
The main contribution adds continuous perceptual tokens to VLM reasoning and makes their representation, supervision and inference roles explicit.
The main deliverable is a new intrinsic surface texture representation with geometric optimization and rendering integration.
The main contribution is a topology representation and generation-to-mesh pipeline that separates continuous prediction from discrete extraction.
The principal contribution changes the diffusion representation and cascaded feature-generation boundary, with targeted photometric and geometric diagnostics.
The main deliverable changes registration from pairwise matching plus graph optimization to conditional joint point generation and rigid recovery.
The main deliverable is a discrete Gaussian-scene representation and structured autoregressive tokenization supporting multiple scene-generation interfaces.
The central deliverable is a deliberately simple global-attention 3D backbone plus training adaptations needed to make it competitive.
The argument primarily changes a loss, supervision signal, optimization/update rule, distillation recipe or adaptation procedure. Some assigned methods update at inference time; the label refers to the objective-driven argument, not only offline training.
Corpus context · 4 papers: median 5.5 main figures and 5.5 main tables; body span ≈ 15.15 pages; attached appendix span ≈ Not resolved pages (0 resolved boundaries). Separate source files: median Not resolved whole-file pages among 0 papers; these can repeat main text and are not appendix-only lengths. Extraction-assisted observations—not drafting targets.
The regularizer or guidance objective is motivated by a diagnosed failure, followed by named variables and controls that test the intended effect. Additional inputs and test-time optimization costs remain explicit.
Editorial starting package: choose artifacts that test your actual claims. Quantities below are planning suggestions, not counts inferred from all category members or conference rules.
One diagram when multiple paths or update stages matter
Mark teachers, students, pseudo-labels, frozen weights, detached gradients and inference-only operations.
When: Use gradient arrows or phase panels when an ordinary module diagram would hide the actual intervention.
An explicit objective; an algorithm box if iterative or stateful
Define variables, positive/negative samples, weights, updates and stopping rather than listing an unexplained weighted sum.
When: A short closed-form loss may need no pseudocode; a multistage update usually does.
One task-specific table or paired baseline/new-rule rows
Hold backbone, supervision and budget fixed where possible so a training-rule claim is not merely a larger-model claim.
When: Show unmatched leaderboard context separately from causal design comparisons.
One appropriate panel per central explanatory claim
Measure the suspected mismatch: gradient behavior, confusing classes, reliability, alignment or temporal persistence.
When: Use the quantity the paper actually claims to improve, not an arbitrary feature visualization.
A compact control table and consequential parameter curves
Distinguish new supervision, loss terms, schedules and extra compute; retain collapse or negative runs.
When: Do not invent large sweeps for inconsequential parameters or hide tuning costs.
Optional
State assumptions and whether the implemented approximation satisfies them.
When: A proof is needed for a formal claim, not as decoration for an empirical improvement.
Exact losses and schedules, coefficients, pseudo-label construction, noise/model mismatch, failed target choices and overhead. Do not hide the only definition of the objective outside the main paper.
Identify frozen estimator and learnable offsets, then show optimization dynamics.
Test changed training and actual inference dependence separately.
Boundary: This route contains 4 primary editorial assignments in the 28-paper inspected subset (28 inventory records). It is not an official track, a universal writing standard or evidence that these choices caused acceptance.
See inline examples of these artifact styles · Compare every member in the measurement ledger
The reusable intervention is test-time physics-guided optimization around frozen normal estimators, not a newly trained backbone.
The main contribution is two complementary latent regularizers motivated by spectral diagnostics and evaluated through diffusion training and reconstruction trade-offs.
The intervention changes how a geometry-aware VLM is trained and fused so that it actually uses geometric evidence, supported by inference-time removal tests.
The argument centers on joint masked semantic learning and probabilistic generative optimization rather than simply introducing another backbone.
A formal characterization, algebraic result, guarantee or theoretically derived explanation organizes the main argument, with algorithms and experiments testing consequences or practical approximations.
Corpus context · 2 papers: median 11 main figures and 0.5 main tables; body span ≈ 14.97 pages; attached appendix span ≈ Not resolved pages (0 resolved boundaries). Separate source files: median Not resolved whole-file pages among 0 papers; these can repeat main text and are not appendix-only lengths. Extraction-assisted observations—not drafting targets.
Both primary theory studies build a physical problem into explicit variables and propositions, then map those factors onto experiments. The small denominator is two and mixes graph sensing and optical statistics.
Editorial starting package: choose artifacts that test your actual claims. Quantities below are planning suggestions, not counts inferred from all category members or conference rules.
One when it makes unfamiliar definitions operational
Connect algebraic symbols to camera geometry, representations or update behavior.
When: Prefer a reusable symbolic example over an unrelated network schematic.
One short map or comparison table when several regimes are involved
Keep quantifiers, non-degeneracy conditions, approximation regimes and guarantee types visible.
When: Especially important when exact theory motivates an approximate practical solver.
As many as the actual logical claims need
State what is proved and identify the central argument before full technical details.
When: No target theorem count; empirical observations must not be promoted to universal guarantees.
Only when there is an empirical consequence to test
Test the predicted difficult regime, scaling, geometry or optimizer behavior under stated assumptions.
When: A finite experiment illustrates or challenges a consequence; it does not prove an asymptotic statement.
Include where needed to prevent a tempting overgeneralization
Explain the role of assumptions and distinguish sufficient from necessary conditions.
When: A theoretical paper need not manufacture a broad benchmark leaderboard if it is irrelevant to the result.
Complete proofs, auxiliary lemmas, derivation details, numerical checks and counterexamples. Essential assumptions and theorem meanings remain in the main manuscript.
Use a small graph to make a connectivity guarantee understandable.
Name equation factors and test the corresponding acquisition choices.
Boundary: This route contains 2 primary editorial assignments in the 28-paper inspected subset (28 inventory records). It is not an official track, a universal writing standard or evidence that these choices caused acceptance.
See inline examples of these artifact styles · Compare every member in the measurement ledger
The central argument links a physical measurement construction to graph connectivity and error control, then validates it in simulation and hardware.
The contribution is a second-order optical theory and recoverability condition, with physical experiments testing consequences rather than replacing the theoretical argument.
The main new knowledge is a controlled measurement, diagnostic, comparison, failure analysis or audit. A proposed remedy, model or tool can follow from that evidence.
Corpus context · 4 papers: median 7 main figures and 2.5 main tables; body span ≈ 14.63 pages; attached appendix span ≈ Not resolved pages (0 resolved boundaries). Separate source files: median Not resolved whole-file pages among 0 papers; these can repeat main text and are not appendix-only lengths. Extraction-assisted observations—not drafting targets.
Interpretation studies separate observable failure, diagnostic access and intervention. Readability, correlation, interpretability and causality are not interchangeable outcomes.
Editorial starting package: choose artifacts that test your actual claims. Quantities below are planning suggestions, not counts inferred from all category members or conference rules.
One compact view
Expose information access, sampling, intervention and matched controls.
When: Essential when teacher labels, dataset splits, task direction or model access determine the interpretation.
One clear entry point
Display the relationship, denominator and variability before a proposed explanation.
When: Show representative ranges rather than selected successes alone.
One destination per central explanatory claim
Distinguish the proposed mechanism from compute, label quality, dataset composition or evaluation artifacts.
When: Correlation and a fitted trend require different language from a controlled causal intervention.
A small aligned set with selection criteria
Make measured failures inspectable and include ambiguous/negative cases where relevant.
When: Qualitative cases illustrate a study; they do not replace prevalence estimates.
As needed by generalization or practicality claims
Repeat the protocol on different tasks/models/settings and report the actual cost.
When: Keep missing, unavailable or diverged results distinct from zero.
Probe configurations, sampling, random controls, dataset strata, additional seeds and failed interpretations. Keep the principal threat model or measurement contract in the main text.
Define random/residual controls before interpreting a low-rank curve.
Group threat-model access levels and preserve metric units.
Boundary: This route contains 4 primary editorial assignments in the 28-paper inspected subset (28 inventory records). It is not an official track, a universal writing standard or evidence that these choices caused acceptance.
See inline examples of these artifact styles · Compare every member in the measurement ledger
The paper's main deliverable is a controlled diagnostic framework for accessible geometric information, not a new general-purpose vision backbone.
The central argument interprets and intervenes in concept circuits, using reconstruction, ablation and model comparisons to assess explanatory utility.
The writing's organizing claim is layer-wise loss of negation readability, with a post-hoc intervention used to test and exploit that diagnosis.
The main contribution exposes and characterizes information leakage under explicitly different access regimes rather than proposing a benign reconstruction system.
A new corpus, acquisition/annotation process, evaluation suite or measurement protocol is the primary reusable deliverable. A companion model demonstrates utility without replacing the need to validate the resource.
Corpus context · 2 papers: median 5.5 main figures and 4.5 main tables; body span ≈ 13.51 pages; attached appendix span ≈ Not resolved pages (0 resolved boundaries). Separate source files: median Not resolved whole-file pages among 0 papers; these can repeat main text and are not appendix-only lengths. Extraction-assisted observations—not drafting targets.
The two primary resource papers explain what new signal or output is measured before introducing a model. Data collection, labeling, metric decomposition and downstream utility perform separate jobs.
Editorial starting package: choose artifacts that test your actual claims. Quantities below are planning suggestions, not counts inferred from all category members or conference rules.
One compact resource table
Compare capabilities, annotation units, acquisition conditions and evaluation tasks, not only total size.
When: Use explicit partial or unavailable coverage rather than forcing binary superiority.
One process diagram with examples
Show raw inputs, transformations, human/machine verification and final annotations.
When: Distinguish generated labels, human corrections and measured ground truth.
A distribution, hierarchy or protocol matrix suited to the labels
Show subjects/scenes/classes, long tails, acquisition diversity and what is held out.
When: Avoid conflating frames, tracks, objects, rendered variants and independent subjects.
Focused tables with the same taxonomy used in the protocol
Demonstrate what the resource reveals and expose non-dominating models or hard subgroups.
When: A high average score alone does not validate label quality or benchmark relevance.
One verification analysis when annotation or metric validity is a claim
Document agreement, quality checks, adversarial controls, contamination checks or calibration.
When: Select the check that addresses this resource's actual failure risks; do not imply a single check proves complete validity.
Capture and synchronization details, annotation instructions, metric computation, release conditions, per-subset results and ambiguity examples. Keep the benchmark's purpose and test unit in the main text.
Define structured outputs and metrics before the model.
Tie new physical supervision to hardware and paired transfer tests.
Boundary: This route contains 2 primary editorial assignments in the 28-paper inspected subset (28 inventory records). It is not an official track, a universal writing standard or evidence that these choices caused acceptance.
See inline examples of these artifact styles · Compare every member in the measurement ledger
The argument first defines a structured video-to-script task, annotation process and hierarchical evaluation, then introduces a model to test that task.
The principal deliverable is synchronized egocentric video and force data, demonstrated through contact supervision and action-transfer experiments.
The central contribution changes execution, inference, memory, acquisition cost or resource scaling while preserving a stated quality or correctness contract.
Corpus context · 5 papers: median 5 main figures and 4 main tables; body span ≈ 15.15 pages; attached appendix span ≈ Not resolved pages (0 resolved boundaries). Separate source files: median Not resolved whole-file pages among 0 papers; these can repeat main text and are not appendix-only lengths. Extraction-assisted observations—not drafting targets.
The system examples separate the changed state or scheduling mechanism from measured deployment costs, and report quality alongside time or memory. They expose workloads and operating regimes where the full configuration is not best.
Editorial starting package: choose artifacts that test your actual claims. Quantities below are planning suggestions, not counts inferred from all category members or conference rules.
One focused breakdown or controlled comparison
Explain which stage is costly and why the proposed decision variable could help.
When: Use matched budgets when an oracle analysis motivates an approximate implementation.
One overview with needed stage or memory insets
Mark preprocessing, cache/reuse, training versus inference, hardware calls and synchronization.
When: The figure should expose added overhead as well as removed work.
One panel per materially different workload
Pair quality with latency, throughput, memory or acquisition effort at comparable operating points.
When: State hardware, precision, batch/sequence/image size and what the timer includes.
A table or stacked breakdown
Show whether an isolated saving survives the entire pipeline.
When: Do not substitute FLOPs, token count or number of steps for wall-clock measurements without qualification.
Curves only for the dimensions the system claims to scale
Measure view/frame count, problem size, cache refresh or device constraints and retain adverse regimes.
When: Changing hardware or input capacity is a different comparison from identical-workload acceleration.
As needed for the execution contract
Make reuse, selection, fallback, stopping and approximation explicit.
When: A claimed exact or distribution-preserving procedure needs a precise argument beyond similar output scores.
Hardware/software configuration, timing boundaries, warmup and repetition, workload distributions, memory definitions, failed optimizations and long-run behavior.
Startup latency, throughput, synchronization and quality require distinct definitions.
Show the evolving attention state and the accuracy/FPS trade-off.
Preserve quality gains from a slower ablation rather than claiming a universal win.
Boundary: This route contains 5 primary editorial assignments in the 28-paper inspected subset (28 inventory records). It is not an official track, a universal writing standard or evidence that these choices caused acceptance.
See inline examples of these artifact styles · Compare every member in the measurement ledger
The central deliverable converts a bidirectional audio-video generator into a streaming system and evaluates startup latency, throughput and fidelity separately.
The writing argues for a simpler retrieval system whose memory construction cost and query-time evidence quality are measured independently.
The contribution is a differentiable rendering framework integrating rational spline parameters with Gaussian sampling and application-specific efficiency tests.
The central design manages streaming geometric state and its cost while retaining long-range reconstruction consistency.
The organizing contribution scales reconstruction context through sparse attention, coarse-to-fine residuals and load-balanced implementation.
The primary argument is a task-specific capability, physical measurement/reconstruction system or domain-grounded method whose validity depends on real acquisition, task assumptions and domain-appropriate evidence.
Corpus context · 1 papers: median 9 main figures and 0 main tables; body span ≈ 15.15 pages; attached appendix span ≈ Not resolved pages (0 resolved boundaries). Separate source files: median Not resolved whole-file pages among 0 papers; these can repeat main text and are not appendix-only lengths. Extraction-assisted observations—not drafting targets.
The single primary application assignment is a computational imaging prototype: physical acquisition, calibration and reconstruction are separately visible. One case is an example, not a measured ECCV-wide application standard.
Editorial starting package: choose artifacts that test your actual claims. Quantities below are planning suggestions, not counts inferred from all category members or conference rules.
One setup or input/output view
Show sensors/modalities, calibration, measured versus generated quantities and what is available at use time.
When: Essential when hardware, multimodal cohorts or nonstandard labels determine the claim.
One overview with necessary physical or anatomical labels
Connect variables, modules and losses to the domain problem rather than only showing generic network boxes.
When: Keep measured inputs, latent physical quantities and predicted outputs visually distinct.
A compact grid or profile plot
Compare matched scenes/patients/objects with relevant maps, spectra, overlays or error views.
When: Label simulation, real data, pseudo-ground-truth and unavailable measurements; an appealing image is not independent validation.
One table per incompatible endpoint
Report the relevant physical/clinical/behavioral metric, units, uncertainty and evaluation split.
When: Separate image quality from geometry, survival discrimination, calibration or downstream behavior.
Controls chosen for the proposed mechanism
Test key physical assumptions, modality use or pipeline components; include an independent reference when feasible.
When: Do not label a model-derived pseudo-label as independent truth or infer clinical benefit from a technical benchmark.
As appropriate to the deployment claim
State calibration, capture, annotation, compute and failure requirements.
When: A successful laboratory demonstration does not establish unrestricted field or clinical reliability.
Calibration, assembly, acquisition settings, reconstruction hyperparameters, repeatability and additional scene failures.
Tie the hardware photograph and ray model to regional image-quality comparisons.
Boundary: This route contains 1 primary editorial assignments in the 28-paper inspected subset (28 inventory records). It is not an official track, a universal writing standard or evidence that these choices caused acceptance.
See inline examples of these artifact styles · Compare every member in the measurement ledger
The main deliverable is a physically realized computational mirror imaging system with a reconstruction operator and real optical demonstrations.
