{"id":"c1f6d925-ea38-4314-87a3-b578e46cae5b","arxiv_id":"2606.09679","paper_version":1,"verdict":"UNVERDICTED","confidence":"LOW","novelty_score":2.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"Four extensions to FOOTPASS baselines for player-centric ball-action spotting in soccer videos achieve 0.548 Macro F1 on test set and 0.446 on challenge set.","lead":"The paper describes four engineering extensions to existing FOOTPASS baselines for detecting which player performs which ball action in soccer broadcast videos. A smart generalist might read it to understand how standard techniques like class reweighting and model fusion address severe imbalance in video action datasets.","discovery_kind":"extension","skeptic_critique":{"model":"grok-4.3","headline":"No significant objection identified","rationale":"The reader's weakest_assumption correctly isolates the dependency on the cited baselines. Because the manuscript offers no contradictory evidence and the claim is a straightforward report of challenge results rather than a theoretical derivation, the assumption does not appear to fail. The UNVERDICTED status is retained solely due to the abstract-only review noted by the reader; no additional load-bearing risk is introduced by the argument itself.","tokens_in":1656,"tokens_out":312,"duration_ms":16208,"concrete_test":"If the authors release code or models, re-execute the full pipeline (including the exact post-processing steps) on the SoccerNet 2026 splits and confirm that the reported Macro F1 values are reproduced within 0.01; if they are not, the contribution of the extensions cannot be isolated.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim is that the four listed extensions to the three FOOTPASS baselines produce a system reaching 0.548 Macro F1 (test) and 0.446 (challenge server). The supporting assumption is that those baselines are a viable starting point and that the extensions (gradient checkpointing, GNN-logit fusion into DST, sqrt-frequency weighting, and the four-stage post-processor) can be applied without new core architectures. No internal inconsistency, hidden assumption in an equation, or unsupported derivation is visible in the provided description; the techniques are standard and the evaluation uses a held-out server set.","agreement_with_reader":"agree"},"referee_report":{"model":"grok-4.3","summary":"The manuscript describes a system for the SoccerNet 2026 Player-Centric Ball-Action Spotting Challenge by extending the three FOOTPASS baselines (TAAD, TAAD+GNN, TAAD+DST) with four modifications: gradient checkpointing for full-backbone fine-tuning, fusion of GNN logits into the DST encoder, square-root frequency class weighting to handle 213:1 class imbalance, and a four-stage post-processing pipeline (per-class logit gating, temporal frame refinement, jersey re-assignment, two-model ensemble). It reports achieving 0.548 Macro F1 on the test set and 0.446 on the challenge set via server evaluation.","tokens_in":1757,"tokens_out":447,"duration_ms":12336,"significance":"If reproducible, the work offers incremental engineering improvements on a challenging player-centric temporal action spotting task with severe class imbalance. The techniques are standard and directly address the stated problem constraints, but the lack of ablations or validation details limits assessment of which extensions drive the reported gains over the cited baselines.","major_comments":[{"comment":"Abstract: the central claims rest on the reported Macro F1 scores (0.548 test, 0.446 challenge) with no accompanying error bars, ablation studies, implementation details, or validation procedure, rendering it impossible to verify that the four listed extensions produce the stated improvements over the FOOTPASS baselines.","section":"Abstract"},{"comment":"No section provides the precise formulation or loss-function integration of the square-root frequency class weighting, despite its identification as a load-bearing extension for the 213:1 imbalance; without this, the contribution cannot be assessed or reproduced.","section":"Methods"}],"minor_comments":[{"comment":"A results table comparing each baseline to the extended system (with and without individual extensions) would clarify the incremental gains.","section":null},{"comment":"The citation to the FOOTPASS baselines [1] should include the exact reference details for the three variants (TAAD, TAAD+GNN, TAAD+DST).","section":null}],"recommendation":"major_revision","confidential_remarks":null},"author_rebuttal":{"model":"grok-4.3","summary":"We thank the referee for the detailed review. We address the two major comments point-by-point below, indicating where revisions will be made to improve clarity and reproducibility.","responses":[{"response":"The manuscript is a concise system-description paper for a fixed challenge deadline rather than a full research article. Implementation details for all four extensions appear in the Methods section. The reported scores are single-run server evaluations on the organizers' fixed test and challenge sets; no error bars are possible without multiple independent runs, which were not performed. Ablation studies were omitted due to the challenge timeline and GPU-hour limits. We will revise the abstract to explicitly state that scores come from single server submissions and to reference the Methods section for extension details.","revision_made":"partial","referee_comment":"[Abstract] Abstract: the central claims rest on the reported Macro F1 scores (0.548 test, 0.446 challenge) with no accompanying error bars, ablation studies, implementation details, or validation procedure, rendering it impossible to verify that the four listed extensions produce the stated improvements over the FOOTPASS baselines."},{"response":"We agree that the current manuscript lacks the explicit formula. In the revised version we will insert the precise definition (square-root inverse-frequency weights applied to the cross-entropy loss) together with the integration equation and the resulting per-class weight values computed from the training-set statistics.","revision_made":"yes","referee_comment":"[Methods] No section provides the precise formulation or loss-function integration of the square-root frequency class weighting, despite its identification as a load-bearing extension for the 213:1 imbalance; without this, the contribution cannot be assessed or reproduced."}],"tokens_in":1333,"tokens_out":373,"duration_ms":16313,"standing_objections":[]},"desk_editor":{"model":"grok-4.3","letter":"This paper extends the three FOOTPASS baselines for player-centric ball-action spotting by adding gradient checkpointing for single-GPU fine-tuning, GNN logit fusion into DST, square-root frequency weighting for the pass-tackle imbalance, and a post-processing chain of gating, temporal refinement, jersey re-assignment, and ensembling. The reported results are 0.548 Macro F1 on the test set and 0.446 on the challenge server.\n\nThe extensions are sensible applications of known techniques. Checkpointing addresses a real hardware constraint, the weighting targets the documented 213:1 skew, and the post-processing steps are the kind of refinements that commonly lift spotting performance. The use of an external server evaluation adds some credibility to the numbers.\n\nThe soft spots are the absence of ablations showing the contribution of each change, no error bars or repeated runs, and high-level descriptions that leave implementation details unclear. The work stays inside the existing baselines rather than adding new core components, so the gains rest on empirical tuning.\n\nThis is mainly for teams already in the SoccerNet challenge or working on broadcast sports video. A reader focused on general action spotting might pick up practical tips on imbalance handling and post-processing, but the scope is narrow.\n\nI would send it to peer review for a challenge or workshop track because the held-out server results are concrete and the methods are described enough to be reproducible by others, even if more analysis would help.","headline":"This is a straightforward engineering report extending FOOTPASS baselines for the SoccerNet 2026 challenge with four standard tweaks, reporting server-evaluated F1 gains but without ablations or variance measures.","tokens_in":2196,"tokens_out":379,"would_cite":false,"duration_ms":17558,"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":"Extensions to FOOTPASS baselines raise Macro F1 to 0.548 on the SoccerNet test set for player-centric ball-action spotting.","keywords":["SoccerNet","ball-action spotting","player-centric","FOOTPASS","post-processing","class weighting","action recognition","video analysis"],"falsifier":"A side-by-side evaluation on the same test set in which the unmodified TAAD+DST baseline alone matches or exceeds 0.548 Macro F1 would show that the extensions add no value.","tokens_in":2550,"feed_emoji":"⚽","tokens_out":715,"duration_ms":16749,"temperature":0.7,"pith_summary":"The paper demonstrates that four targeted changes to three existing baseline models can improve results on a broadcast soccer video task that requires identifying which player performs which of eight actions and at what exact moment. The changes include enabling full model retraining on limited hardware, combining graph-based and visual features, reweighting rare actions to counter extreme imbalance, and applying a sequence of prediction cleanup steps plus an ensemble. A reader would care because reliable automatic detection of player actions supports downstream uses such as game statistics, scouting, and broadcast enhancement. The work treats the baselines as a workable starting point and shows measurable gains from the listed additions rather than a complete redesign.","feed_headline":"Extensions lift soccer spotting Macro F1 to 0.548","feed_subtitle":"Four changes to existing baselines improve detection of who performs which action and when in broadcast soccer.","key_machinery":"The four-part extension pipeline (gradient checkpointing, GNN-to-DST logit fusion, square-root class weighting, and multi-step post-processing with ensemble) applied to the three FOOTPASS baselines.","core_discovery":"By applying gradient checkpointing to permit full-backbone fine-tuning, fusing GNN logits into the DST encoder, adopting square-root frequency class weighting, and running a post-processing pipeline of per-class logit gating, temporal frame refinement, jersey re-assignment, and a two-model ensemble on the TAAD, TAAD+GNN, and TAAD+DST baselines, the system reaches 0.548 Macro F1 on the test set and 0.446 on the challenge set.","pith_inferences":["The same four extensions could be tested on other video action datasets that exhibit similar class imbalance.","The post-processing pipeline might be applied independently to outputs from entirely different spotting models to measure its isolated contribution.","An expanded ensemble that includes additional variants of the baselines could be evaluated to check whether further gains remain available.","The reported scores provide a new reference point for future submissions that wish to compare against these particular extensions rather than the raw baselines."],"forward_implications":["Gradient checkpointing makes full fine-tuning of large visual backbones feasible on a single GPU.","Fusing GNN logits into the DST encoder adds tactical graph context to per-player visual features.","Square-root frequency weighting reduces the dominance of frequent classes such as passes over rare ones such as tackles.","The post-processing steps correct timing errors, re-assign players via jersey numbers, and combine two models to raise final accuracy."],"fun_headline_variants":["Extensions to FOOTPASS baselines reach 0.548 Macro F1 in SoccerNet 2026","Player-centric soccer spotting reaches 0.548 Macro F1 via four extensions","TAAD baselines updated to achieve 0.548 Macro F1 on SoccerNet 2026","SoccerNet 2026 reaches 0.548 Macro F1 with FOOTPASS extensions","Four extensions yield 0.548 Macro F1 for SoccerNet player spotting"],"cache_read_input_tokens":2112,"weakest_assumption_plain":"The three FOOTPASS baselines already supply a workable foundation that the four listed extensions can improve without new core model architectures.","fun_headline_variants_meta":{"raw":{"variants":["Extensions to FOOTPASS baselines reach 0.548 Macro F1 in SoccerNet 2026","Player-centric soccer spotting reaches 0.548 Macro F1 via four extensions","TAAD baselines updated to achieve 0.548 Macro F1 on SoccerNet 2026","SoccerNet 2026 reaches 0.548 Macro F1 with FOOTPASS extensions","Four extensions yield 0.548 Macro F1 for SoccerNet player spotting"]},"model":"grok-4.3","cost_usd":0.006965,"raw_usage":{"total_tokens":3218,"prompt_tokens":649,"num_sources_used":0,"completion_tokens":112,"cost_in_usd_ticks":69649500,"prompt_tokens_details":{"text_tokens":649,"audio_tokens":0,"image_tokens":0,"cached_tokens":256},"completion_tokens_details":{"audio_tokens":0,"reasoning_tokens":2457,"accepted_prediction_tokens":0,"rejected_prediction_tokens":0}},"tokens_in":649,"tokens_out":112,"duration_ms":14895,"temperature":1.0,"reasoning_tokens":2457,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-06-27T17:19:19.092530+00:00","model_set":{"reader":"grok-4.3"},"falsifier":"A side-by-side evaluation on the same test set in which the unmodified TAAD+DST baseline alone matches or exceeds 0.548 Macro F1 would show that the extensions add no value.","supporting_citations":[],"review_version":1}