{"id":"ac32d46b-a73f-40b6-ac4e-ee89f7810116","arxiv_id":"2508.02288","paper_version":1,"verdict":"UNVERDICTED","confidence":"LOW","novelty_score":5.0,"correctness_risk":"unknown","formal_verification":"none","parameter_count":0,"one_line_summary":"A stereo event-camera-only framework with a dual filter and object-centric box alignment outperforms synchronized-sensor continuous-time 3D detectors in dynamic scenes.","lead":"The authors built a 3D object detector that uses only stereo event cameras, which record changes in a scene instead of fixed frames. In fast-motion tests it beat prior hybrid approaches that still rely on LiDAR or regular cameras, pointing toward cheaper continuous-time perception.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The central claim rests on experiments in a body text that is unreadable in this submission, so the event-only stereo premise cannot be checked; independent evidence from the released code is required before a verdict.","rationale":"The reader's verdict is UNVERDICTED because only the abstract is readable; I agree. The single most load-bearing concern is the same as the reader's weakest assumption: the abstract's claim that event-only stereo data can support 3D detection after a dual filter. That claim is entirely empirical, and the body of the paper is the only place where it is supported. Because the body text is corrupted in the provided input, the experimental evidence cannot be inspected. This is not a manufactured concern; the unreadable text itself contains a visible artifact from a different arXiv paper, confirming that the rendering is broken rather than simply dense. I considered declaring 'no significant objection,' but that would be misleading: the absence of readable evidence is itself the objection to accepting the paper. The reader already reached UNVERDICTED, and my analysis does not move that verdict, hence UNCHANGED. A concrete test is proposed: recover the original source from arXiv or run the released code on a benchmark. If the code reproduces the claimed numbers, the concern is resolved; if not, the paper should remain unverdictable until the evidence is provided. I have not questioned author integrity, only the inspectability of the evidence, which is a legitimate scientific concern.","tokens_in":6915,"tokens_out":2668,"duration_ms":33566,"concrete_test":"Download the arXiv source files for 2508.02288 and render them properly to recover the body text; then inspect the experiments section for (a) the dataset (e.g., DSEC, MVSEC, or custom), (b) whether the 'prior approaches' baseline includes an event-plus-RGB/LiDAR fusion method evaluated on the same dynamic sequences, and (c) metrics such as 3D AP and translation/rotation errors with variance or error bars. If the source is genuinely corrupted on arXiv, instead clone https://github.com/mickeykang16/Ev-Stereo3D and run it on the DSEC high-speed split, comparing the reproduced 3D AP to the values claimed in the abstract. If neither the recovered paper nor the code yields reproducible numbers, the central claim remains unverified and the correct verdict stays UNVERDICTED.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The strongest claim is empirical: a dual filter plus object-centric regression turns raw stereo event streams into state-of-the-art continuous-time 3D detection without LiDAR or RGB. For this to hold, (1) the event stream must contain enough semantic and geometric signal for 3D box regression after filtering, and (2) the experiments must be fair and reproducible. The provided full text is corrupted mojibake; it even contains the header of an unrelated math.CT paper (arXiv:2508.02285), so no method details, equations, tables, or ablations can be inspected. This is not an internal inconsistency but an evidence gap: the abstract itself concedes that event data lack semantic and geometric information, and the body is the only place where the dual filter's compensation is validated. No machine-checked proofs, reproducible code artifacts, or parameter-free derivations are available in the readable portion. The GitHub link exists, but the abstract provides no commit hash or reproducibility statement, so the code cannot be treated as independent verification. Consequently, the pivotal empirical premise is unsupported in the available record, making the paper unverdictable rather than accepted or rejected.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript proposes an event-only stereo 3D object detection framework that relies solely on event cameras, eliminating the need for conventional 3D sensors such as LiDAR or RGB cameras. The proposed method introduces a dual filter mechanism to extract both semantic and geometric information from event data, and an object-centric bounding-box regression strategy. The abstract claims that the method outperforms prior approaches in dynamic environments and provides a GitHub repository link. However, the full text of the submission is severely corrupted and unreadable, and the abstract contains no quantitative results, baselines, dataset statistics, or evaluation metrics. As a result, the central claim cannot be verified from the available record.","tokens_in":7138,"tokens_out":4507,"duration_ms":56375,"significance":"If the claimed results hold, the contribution would be significant for event-based vision and continuous-time 3D perception: an event-only stereo detector that avoids synchronization with conventional sensors could address high-speed scenarios where frame-based systems suffer. The paper's architectural idea (dual filter plus object-centric regression) and the public code link are positive aspects. However, because the body text is corrupted and the abstract lacks any numbers, the scientific content cannot be assessed. The contribution therefore remains unverified in this submission, and its significance is conditional on a complete and readable manuscript.","major_comments":[{"comment":"The body of the manuscript is corrupted mojibake and includes the header of an unrelated arXiv paper (arXiv:2508.02285 [math.CT]). None of the method description, equations, architecture details, training procedures, experimental setups, or result tables can be read. This makes it impossible to check any aspect of the central claim that the proposed method outperforms prior approaches, and the paper is not in a reviewable state.","section":"Full Text"},{"comment":"The key empirical claim that the method 'outperforms prior approaches in dynamic environments' is not supported by any quantitative information: no mAP or other metric, no baseline names, no dataset description, and no error bars are provided. As written, this is an assertion rather than a demonstrated result, and it cannot be evaluated or reproduced from the abstract alone.","section":"Abstract"},{"comment":"The abstract motivates the dual filter by stating that event data lack semantic and geometric information, but the only place where this compensation mechanism is validated is the unreadable body text. The efficacy of the dual filter is therefore a load-bearing dependency of the central claim, and the available record provides no evidence that it works.","section":"Abstract and Full Text"}],"minor_comments":[{"comment":"The GitHub repository link should include a specific commit hash, a license, and a reproducibility statement, so that readers can verify the code version that produced the reported experimental results.","section":"Abstract"},{"comment":"Even after the full text is repaired, the authors should ensure that the abstract reports concrete quantitative results (for example, AP on a standard event-based 3D detection benchmark) rather than only a qualitative superiority claim.","section":"Abstract"}],"recommendation":"major_revision","confidential_remarks":"The submitted full text is corrupted beyond readability, so I could not perform a technical review. I recommend that the editor return the manuscript to the authors with a request to resubmit a cleanly compiled PDF, after which a full review can be conducted. The major revision recommendation reflects the need for a readable and complete manuscript; it is not an assessment of the scientific merit, which is currently impossible to judge."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: this is a paper about making 3D object detection work with only stereo event cameras, removing the need for synchronized LiDAR/RGB. That's a genuinely interesting direction—the abstract's claim to beat hybrid approaches in fast-motion scenes would matter for autonomous perception. The idea of a dual filter for semantic/geometric cues and object-centric box regression is plausible.\n\nWhat the paper does well: the framing is clear and honest. It acknowledges that event data lacks semantic and geometric information, which is exactly the hard part. The authors also provide a code link, which helps if the code is actually usable.\n\nThe problem: the full text is unreadable mojibake. It even contains the header of an unrelated math.CT paper. I can't inspect the architecture, the equations, the datasets, the ablations, or the numbers. The abstract gives no quantitative results, no baselines, no error bars. So the central empirical claim—that this method outperforms prior work in dynamic environments—rests entirely on experiments I cannot see. The stress-test note is right: this is an evidence gap, not an internal contradiction. The abstract itself concedes the key difficulty, and the body is the only place where the dual filter's compensation is validated.\n\nThe reader's scores seem about right. Significance-if-true is moderate; novelty is plausible; soundness is low only because there is nothing to check. I'd add that the GitHub link without a commit hash isn't independent verification.\n\nBottom line: I can't recommend reviewing this submission as-is. The right move is to send it back to the authors to fix the PDF and resubmit with a complete, readable text. If the clean version matches the abstract's promises, it deserves a serious referee. On the current evidence, it's unverdictable, not wrong.","headline":"Event-only stereo 3D detection is worth a look, but the mangled full text makes the paper unverdictable; fix the PDF and then review it.","tokens_in":7595,"tokens_out":2123,"would_cite":false,"duration_ms":26445,"reading_group":"no","serious_thinker":"unclear","would_accept_peer_review":false},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Stereo event cameras alone can detect 3D objects continuously, without LiDAR or RGB frames, using a dual filter for semantic and geometric cues and object-centric box regression.","keywords":["event cameras","stereo vision","3D object detection","continuous-time perception","asynchronous sensors","autonomous driving","dual filter mechanism","object-centric regression"],"falsifier":"Take a slowly moving, low-texture scene where a LiDAR-based detector still finds objects but event cameras fire only a handful of events per second; if the proposed detector's 3D boxes collapse or vanish under those conditions, the claim that event streams alone suffice for continuous-time detection fails.","tokens_in":6756,"feed_emoji":"🚗","tokens_out":4546,"duration_ms":48148,"temperature":0.7,"pith_summary":"3D object detection usually depends on LiDAR or RGB cameras, whose fixed frame rates miss fast motion. The paper proposes a detection framework that uses only a stereo pair of event cameras, asynchronous sensors that fire when brightness changes, so detection can happen continuously in time. To make up for the sparse semantic and geometric content of raw events, it introduces a dual filter that separates the two kinds of information, and it regresses 3D boxes in an object-centric way. The authors report that this event-only method outperforms prior approaches in dynamic environments, arguing that event cameras alone can support reliable 3D perception.","feed_headline":"No LiDAR, no RGB: stereo events detect 3D objects continuously","feed_subtitle":"Dual filtering pulls semantic and geometric cues from raw event streams to close high-speed perception gaps.","key_machinery":"The load-bearing mechanism is the dual filter applied to raw stereo event streams. Event cameras are asynchronous sensors that emit a pulse whenever pixel brightness changes, giving high temporal resolution but no direct color or depth. The dual filter splits those events into two representations: one tuned for semantic cues such as object boundaries and one tuned for geometric cues such as depth and shape. The detector then predicts 3D boxes in an object-centric coordinate frame, meaning each box is regressed relative to the object's own center rather than a fixed global frame, which the authors use to make regression more stable under motion.","core_discovery":"The central claim is that a stereo event-camera setup, with no LiDAR and no RGB frames, can serve as a complete 3D object detector for dynamic scenes. The authors argue that the usual objection to event data, that it lacks the semantic and geometric richness needed for box regression, can be overcome by a dual filter that extracts both types of information, followed by object-centric bounding-box regression. On their experiments, the framework outperforms prior approaches in dynamic environments, which supports the conclusion that continuous-time 3D detection does not require synchronized conventional sensors.","pith_inferences":["Beyond the paper, the same dual-filter separation could be tested on other continuous-time perception tasks, such as object tracking, depth completion, or drivable-space estimation from stereo events.","A direct but untested consequence is that the detector's reliance on motion means performance should degrade on static or low-texture scenes; a benchmark that varies event rate and texture would map where the method's advantage ends.","Beyond the paper, the object-centric regression head might also make the detector more robust to changes in camera pose or vehicle motion than global-frame predictors, a claim that could be checked by cross-sequence generalization tests."],"forward_implications":["If the central claim holds, autonomous vehicles can perceive fast-moving objects with sensor streams that never skip a frame, eliminating the perception gap caused by fixed frame rates.","Event-only detection removes the need for synchronized LiDAR or RGB input, simplifying sensor setups and avoiding calibration drift between modalities.","Continuous-time detections could be passed directly to planning and control at event rate rather than frame rate, enabling faster reaction to sudden motion.","The dual-filter result demonstrates that sparse asynchronous event data can be separated into semantic and geometric streams, offering a path for other geometry-and-semantics perception tasks."],"supporting_citations":[],"fun_headline_variants":["Stereo event cameras alone enable continuous 3D detection","No LiDAR, no RGB: event streams power 3D detection","Event-only stereo detects 3D objects in fast motion","Continuous-time 3D detection with just stereo events","Dual filters turn event data into 3D box regressions"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that raw stereo event streams, once filtered, contain enough semantic and geometric information to regress precise 3D bounding boxes; if a scene produces too few events or too little texture, the detector has no signal to work with.","fun_headline_variants_meta":{"raw":{"variants":["Stereo event cameras alone enable continuous 3D detection","No LiDAR, no RGB: event streams power 3D detection","Event-only stereo detects 3D objects in fast motion","Continuous-time 3D detection with just stereo events","Dual filters turn event data into 3D box regressions"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000234,"raw_usage":{"total_tokens":1444,"prompt_tokens":838,"completion_tokens":606,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":454,"completion_tokens_details":{"reasoning_tokens":521}},"tokens_in":454,"tokens_out":606,"duration_ms":6195,"temperature":1.0,"reasoning_tokens":521,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T05:01:14.960718+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a slowly moving, low-texture scene where a LiDAR-based detector still finds objects but event cameras fire only a handful of events per second; if the proposed detector's 3D boxes collapse or vanish under those conditions, the claim that event streams alone suffice for continuous-time detection fails.","supporting_citations":[],"review_version":1}