{"id":"8bb4f900-ccc3-48d9-b68f-e3ff3a38b50a","arxiv_id":"2605.24074","paper_version":1,"verdict":"UNVERDICTED","confidence":"LOW","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"WideDepth supplies the first millimeter-accurate indoor fisheye depth benchmark together with a stereo generation pipeline and model adaptation technique.","lead":"The paper introduces WideDepth, the first indoor fisheye depth dataset with 101 scenes, 5K stereo pairs, and millimeter-level ground truth derived from LiDAR. It also offers an adaptation method for pinhole-trained models and extra sparse depth samples for fine-tuning.","discovery_kind":"new_application","skeptic_critique":{"model":"grok-4.3","headline":"Millimeter GT accuracy for fisheye views rests on the unvalidated assumption that LiDAR-to-image projection introduces no systematic bias from calibration or distortion modeling.","rationale":"The reader's weakest_assumption correctly isolates the single assumption that must hold for the dataset's central selling point (millimeter-accurate fisheye GT) to be credible. Full-text review does not remove the concern because the accuracy validation step remains the missing link; the rest of the paper (model adaptation, evaluations) is downstream of this claim.","tokens_in":1717,"tokens_out":374,"duration_ms":24005,"concrete_test":"Locate the methods subsection on LiDAR labeling and fisheye projection; if it contains no quantitative accuracy experiment (RMSE or max error on known-distance targets across the full FOV), re-run the pipeline on a synthetic scene with known ground-truth geometry and measure the output depth error at 1 m and 3 m distances.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The headline claim requires that the high-resolution LiDAR scans yield true millimeter-level depth and disparity labels after projection into all fisheye (and stereo) views. This holds only if (a) extrinsic calibration between LiDAR and each fisheye camera is accurate to sub-millimeter level, (b) the fisheye distortion model used for projection matches the physical lens to within the claimed tolerance, and (c) occlusions and surface reflectance do not produce holes or erroneous depths that are then treated as ground truth. The paper describes a LiDAR-based generation pipeline and states millimeter accuracy, yet provides no independent validation (e.g., comparison against a second metrology instrument on calibration targets or error histograms stratified by range and angle). Without that check, the accuracy claim is an assertion rather than a demonstrated result.","agreement_with_reader":"agree"},"referee_report":{"model":"grok-4.3","summary":"The manuscript introduces WideDepth as the first indoor fisheye depth estimation benchmark, comprising 101 scenes with 5K high-resolution stereo pairs that include millimeter-level ground truth depth and disparity labels generated via a high-resolution LiDAR scanning and projection pipeline. It provides paired pinhole and fisheye imagery across multiple fields of view and stereo baselines, proposes an adaptation method for pinhole-trained stereo models to fisheye data, describes a LiDAR-based stereo fisheye image generation pipeline, evaluates state-of-the-art monocular, stereo, and depth completion models, and supplies 18K sparse depth samples that reportedly yield up to a 62% performance boost when fine-tuning pinhole models on fisheye data.","tokens_in":1879,"tokens_out":411,"duration_ms":22891,"significance":"If the millimeter-level ground truth accuracy can be independently validated, the dataset would address a clear gap by supplying the first dedicated high-precision indoor benchmark for fisheye depth estimation, directly supporting robotics applications that rely on fisheye cameras for near-field manipulation and navigation.","major_comments":[{"comment":"Abstract: the central claim of 'millimeter-level ground truth depth and disparity' for all fisheye views rests on an unvalidated LiDAR-to-image projection pipeline; no error histograms, calibration target comparisons, or stratified accuracy metrics (by range or angle) are supplied to confirm sub-millimeter extrinsic calibration and distortion model fidelity, which directly undermines the title and dataset utility.","section":"Abstract"},{"comment":"Abstract: the reported 'up to a 62% performance boost' from fine-tuning with the 18K LiDAR-derived sparse depth samples is presented without baseline values, absolute error metrics, or per-model breakdowns, rendering the quantitative claim impossible to evaluate against the stated evaluation of SOTA models.","section":"Abstract"}],"minor_comments":[],"recommendation":"major_revision","confidential_remarks":null},"author_rebuttal":{"model":"grok-4.3","summary":"We thank the referee for the constructive feedback. We address each major comment below and will revise the manuscript accordingly to strengthen the presentation of our claims.","responses":[{"response":"We agree that the current manuscript does not supply the requested validation artifacts (error histograms, calibration target comparisons, or stratified metrics) to fully substantiate the millimeter-level claim across fisheye views. The LiDAR projection pipeline is described in Section 3, but explicit quantitative validation of extrinsic calibration and distortion fidelity is missing. We will add these analyses in the revised version, including error distributions, target-based comparisons, and breakdowns by range and angle, to support the title and dataset claims.","revision_made":"yes","referee_comment":"[Abstract] Abstract: the central claim of 'millimeter-level ground truth depth and disparity' for all fisheye views rests on an unvalidated LiDAR-to-image projection pipeline; no error histograms, calibration target comparisons, or stratified accuracy metrics (by range or angle) are supplied to confirm sub-millimeter extrinsic calibration and distortion model fidelity, which directly undermines the title and dataset utility."},{"response":"We acknowledge that the abstract presents the 62% figure without accompanying baseline values, absolute metrics, or per-model details, which prevents direct evaluation. This result originates from the fine-tuning experiments in Section 5.3 comparing adapted models to their pinhole baselines on fisheye data. In the revision we will expand both the abstract and the experimental section to report the necessary baselines, absolute errors, and per-model breakdowns so the claim can be assessed alongside the SOTA evaluations.","revision_made":"yes","referee_comment":"[Abstract] Abstract: the reported 'up to a 62% performance boost' from fine-tuning with the 18K LiDAR-derived sparse depth samples is presented without baseline values, absolute error metrics, or per-model breakdowns, rendering the quantitative claim impossible to evaluate against the stated evaluation of SOTA models."}],"tokens_in":1396,"tokens_out":430,"duration_ms":29350,"standing_objections":[]},"desk_editor":{"model":"grok-4.3","letter":"The main thing here is a new indoor fisheye depth dataset called WideDepth with 101 scenes, 5K stereo pairs, paired pinhole images, varying baselines, and a LiDAR-based generation pipeline. They also describe adapting pinhole stereo models to fisheye data and report a 62% boost from fine-tuning with their 18K sparse samples, plus evaluations of monocular, stereo, and completion models.\n\nThe paper does a straightforward job filling a documented gap. Fisheye cameras see use in near-field robotics, but indoor benchmarks with accurate depth labels for them have been missing. Supplying both dense and sparse data across horizontal and vertical setups gives practical material for testing transfer and stereo matching.\n\nThe soft spot is the soundness of the core accuracy claim. The abstract asserts millimeter-level ground truth from LiDAR scans but supplies no validation metrics, calibration details, error histograms, or checks against independent instruments. The stress-test point about possible systematic bias from extrinsic calibration, distortion modeling, occlusions, or surface properties lands directly on the abstract. Without those checks shown, the millimeter figure is an assertion rather than a demonstrated result. If the full methods section has independent verification, that would fix it; otherwise it is the load-bearing weakness.\n\nThis is for researchers in robotics perception and fisheye depth estimation who need new indoor data. A reader working on model adaptation or stereo for wide-angle cameras would get usable baselines and samples from it. It deserves a serious referee so the pipeline and accuracy can be examined in detail.","headline":"WideDepth brings a new fisheye depth dataset with LiDAR labels and an adaptation method, but the millimeter accuracy claim has no visible validation or error analysis.","tokens_in":2353,"tokens_out":392,"would_cite":false,"duration_ms":42041,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"grok-4.3","headline":"WideDepth supplies the first indoor fisheye depth benchmark with millimeter-accurate LiDAR labels.","keywords":["fisheye depth estimation","indoor benchmark","LiDAR ground truth","stereo dataset","robotics perception","depth completion","model adaptation","millimeter accuracy"],"falsifier":"Independent high-precision measurements on the same scenes that show average depth label errors larger than a few millimeters.","tokens_in":2640,"feed_emoji":"📷","tokens_out":775,"duration_ms":37202,"temperature":0.7,"pith_summary":"The paper introduces WideDepth as the first indoor dataset for fisheye depth estimation in robotics settings. It contains 101 scenes with 5,000 high-resolution stereo pairs carrying millimeter-level ground truth depth and disparity, plus paired pinhole images and multiple stereo baselines and orientations. The authors also describe a LiDAR-based pipeline to generate the fisheye images and a method for adapting pinhole-trained models to fisheye data. They evaluate current monocular, stereo, and depth-completion models on the benchmark and show that fine-tuning with 18,000 additional sparse LiDAR samples improves performance by up to 62 percent.","feed_headline":"Fisheye depth benchmark reaches millimeter accuracy with LiDAR","feed_subtitle":"WideDepth supplies 5K stereo pairs from 101 indoor scenes plus an adaptation method and 18K sparse samples for robotics models.","key_machinery":"The WideDepth dataset together with its LiDAR-based stereo fisheye image generation and labeling pipeline that supplies millimeter-accurate ground truth.","core_discovery":"WideDepth is the first indoor dataset for fisheye depth estimation, featuring 101 scenes containing 5K high-resolution stereo pairs labeled with millimeter-level ground truth depth and disparity. The dataset also includes paired pinhole and fisheye samples across varying fields of view and baselines in both horizontal and vertical stereo setups. A method to adapt pinhole-trained stereo models to fisheye images is proposed together with a novel stereo fisheye image generation pipeline based on high-resolution LiDAR scans. State-of-the-art models are evaluated on the benchmark, and 18K LiDAR-derived sparse depth training samples are released that achieve up to a 62 percent performance boost wh","pith_inferences":["The dataset could support development of fisheye-specific network architectures instead of relying only on adaptation from pinhole models.","Millimeter accuracy enables testing of depth methods for near-field robotic manipulation tasks where centimeter-scale errors are unacceptable.","The LiDAR-to-fisheye generation pipeline could be reused to create training data for other wide-angle camera models or outdoor settings.","Direct comparisons across the dataset's horizontal and vertical stereo setups could identify configurations that minimize distortion effects."],"forward_implications":["State-of-the-art monocular, stereo matching, and depth completion models can be evaluated on fisheye data with precise quantitative metrics.","Pinhole-trained stereo models can be adapted to fisheye images using the supplied adaptation method and paired samples.","Fine-tuning pinhole-based models with the 18K LiDAR-derived sparse depth samples yields up to 62 percent performance improvement on fisheye data.","The benchmark supports research on varying fields of view, baselines, and horizontal versus vertical stereo configurations for indoor robotics."],"fun_headline_variants":["Fisheye depth benchmark at millimeter accuracy","Millimeter-accurate dataset for fisheye depth","WideDepth benchmark for indoor fisheye depth estimation","Stereo fisheye depth with LiDAR millimeter ground truth","Millimeter ground truth for indoor fisheye stereo pairs"],"cache_read_input_tokens":2112,"weakest_assumption_plain":"The LiDAR scanning and labeling pipeline produces true millimeter-level accuracy for all fisheye views and stereo configurations without systematic bias from calibration, occlusion, or surface properties.","fun_headline_variants_meta":{"raw":{"variants":["Fisheye depth benchmark at millimeter accuracy","Millimeter-accurate dataset for fisheye depth","WideDepth benchmark for indoor fisheye depth estimation","Stereo fisheye depth with LiDAR millimeter ground truth","Millimeter ground truth for indoor fisheye stereo pairs"]},"model":"grok-4.3","cost_usd":0.008927,"raw_usage":{"total_tokens":4047,"prompt_tokens":737,"num_sources_used":0,"completion_tokens":67,"cost_in_usd_ticks":89274500,"prompt_tokens_details":{"text_tokens":737,"audio_tokens":0,"image_tokens":0,"cached_tokens":256},"completion_tokens_details":{"audio_tokens":0,"reasoning_tokens":3243,"accepted_prediction_tokens":0,"rejected_prediction_tokens":0}},"tokens_in":737,"tokens_out":67,"duration_ms":40561,"temperature":1.0,"reasoning_tokens":3243,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-06-30T15:41:17.192018+00:00","model_set":{"reader":"grok-4.3"},"falsifier":"Independent high-precision measurements on the same scenes that show average depth label errors larger than a few millimeters.","supporting_citations":[],"review_version":1}