Computer Vision and Pattern Recognition · arXiv:2611.16408 v2 · submitted 12 Nov 2026
Gravity and Flight: Learning to Navigate Fine-Grained Remote Sensing Scenes
The paper says
“Vision-language models are increasingly capable, yet their reasoning in complex remote sensing scenes can remain coarse-grained and insufficiently grounded in task-relevant visual evidence.”
more of the abstract
We propose Gravity and Flight (G&F), a framework for scene specialization with limited supervision that separates visual and reasoning verification while coupling policy updates. It combines Evidence-of-Thought supervised fine-tuning (EoT-SFT) and Relational Transfer Policy Optimization (RTPO). EoT-SFT teaches fine-grained observation and faithful reasoning through description, analysis, and supported conclusions. RTPO independently samples responses to scene-matched image pairs with fine-grained visual differences under a shared task. Its rubric-based verification combines Gravity, which checks visual fidelity and description coverage, and Flight, which takes descriptions as premises to assess evidence use and relations between paired judgments. Their scores and task correctness yield a shared pair-level advantage that couples policy updates from both responses. Using 1,008 demonstrations and 500 training pairs for harbor specialization, G&F achieves a mean score of 69.12 across six evaluation sets, exceeding the backbone and supervised initialization by 9.76 and 4.56 points. Matched controls show gains over single-image and joint-input RL, and removing either verification view reduces performance. Applying G&F to urban specialization also improves on the base model and supervised initialization across three evaluation subsets. Project details and updates are available at https://github.com/Z0ngL1ng/Gravity-and-Flight.
from the abstract
Pith says
Machine review · deepseek-v4-flash · 13 Nov 2026 · of version 1 of the paper; version 2 is newer · what was read was not recorded · not checked by a person
It can be wrong. Challenge this review or reply to it.
The passes disagreed. Reader: holds up. Skeptic: agrees with the reader. Referee: major revision.
Solid empirical paper with a clean method and honest limitations, but the single-run evaluation leaves the headline improvement unaccompanied by any variance estimate. This line is wrong
2 major objections · 6 minor
How often objections from reviews like this one are right: not yet measured
Not checked: figures, code, what the references say.
The paper says, at Section G, sentence b615
The review did not quote this passage.
Pith says: Objection 1 of 2 · major
The paper reports a single run per experimental setting. RTPO training involves multiple stochastic sources: (1) random response sampling at temperature 0.6, (2) random one-to-one matching of responses into candidate pairs, and (3) verifier-score variability (F2 repeat-change rate: 29.17%, Table S4).
more of objection 1
The 4.56-point gain over M1—a central claim—is not accompanied by any variance estimate. Without multiple seeds or bootstrapped confidence intervals, it is unknown whether this gain is statistically robust. The paper acknowledges the limitation only for cost differences (b615: 'differences below a few percent should be read as run-to-run variation rather than as systematic cost differences'), not for performance. This is load-bearing because the headline numeric result could shift with run-to-run variation.
The paper says, at Table S4 and Section C.2
The review did not quote this passage.
A classifier read this sentence of the paper as acknowledging the concern (probability 0.71):
“Such variability can perturb the relative rewards used to compute the advantage.”
Pith says: Objection 2 of 2 · major
The verifier reliability study reports F2 (paired adaptation) agreement within one point of human ratings at 83.33% and a repeat-change rate of 29.17% on 24 units. The paper acknowledges (b475) that '24 units support an indication of reliability rather than a precise estimate' and (b472) that variability 'can perturb the relative rewards used to compute the advantage.'
more of objection 2
However, the paper does not bound how this verifier noise propagates to the final evaluation scores or the policy update. Since the reward signal is central to RTPO, the sensitivity of the reported gains to verifier noise should be assessed, e.g., by re-running the evaluation with perturbed scores or by reporting variance across verifier calls.
6 minor objections
The paper says, at Table 1
Pith says: minor objection 1 of 6
The 'Data Scale' column aggregates different types of supervision (pretraining, instruction tuning, paired training) into a single number. For G&F, 2K items includes both demonstrations and pairs, while for GeoChat (318K) the count includes full instruction-tuning data. Clarify what each count covers so that the 'limited supervision' claim is interpretable as an apples-to-apples comparison.
The paper says, at Table 3, urban results
Pith says: minor objection 2 of 6
The base-model scores on EarthVQA Urban (11.00) and RSVQA HR (15.00) are very low, which makes the relative gains (+20.43 and +28.67 after +RTPO) large in absolute terms but also raises the question of whether the base model is fundamentally unsuited to the urban evaluation protocols. Report whether the evaluation metrics are the same as those used for harbor, and confirm that the base model's low scores are not an artifact of formatting or answer extraction.
show 4 more
The paper says, at Section 4.2, Table 1
Pith says: minor objection 3 of 6
The comparison to general VLMs (LLaVA-1.5, Qwen3.5) is informative but the data-scale gap makes it hard to separate domain specialization from base-model capability. Consider adding a column for domain-specific vs. general model status, or note explicitly that G&F fine-tunes a 4B model while several baselines use 7B parameters.
The paper says, at Table S2, line 'Policy' entry
Pith says: minor objection 4 of 6
The policy's available reference inputs are listed as 'Own image and shared question with public conditions.' This is stated elsewhere but the table would benefit from a note that the public conditions are part of the shared instruction q.
The paper says, at Section G, cost reporting
Pith says: minor objection 5 of 6
The cost measurements (Table 2) are wall-clock time on a specific two-GPU machine with online-verifier waiting. While this is informative, the practical deployment cost is dominated by verifier API calls. Report the approximate number of verifier calls per setting (27,000 per pass is stated in Section D.2; this could be moved into Section G or Table 2).
The paper says, at General
Pith says: minor objection 6 of 6
Minor typographical issues: line 'relational reasoning' is hyphenated inconsistently; 'G&F achieves a higher mean score than the backbone' could be 'G&F achieves higher mean scores than the backbone' for grammatical agreement. These do not affect clarity.
What would settle it
Run the full harbor RTPO training with five different random seeds, each producing a final model. Evaluate all models on the same six evaluation sets and report the mean and standard deviation of the overall score. If the mean improvement over M1 is at least 2 points and the standard deviation is less than 1 point, the claim is robust; otherwise the improvement may not be reliably attributable to the method.
from the skeptic's reading
How the review went
4 readings by deepseek-v4-flash · 13 Nov 2026
- Reader holds up
- Skeptic partly agrees
- Editor wrote the verdict above
- Referee major revision: 2 major objections, 6 minor
Not run on this paper: the simulated authors' defence.
Pith did not check this review's objections against the paper.
The review in full
The review in brief, machine-written
This is a well-built paper that does what it says on the tin. The two-stage design—EoT-SFT to teach structured observation, then RTPO to refine on scene-matched pairs with decoupled visual and reasoning verifiers—is clearly motivated and carefully ablated. The authors construct their own supervision data (1,008 demos, 500 pairs) with explicit Evidence and Boundary Cards, run matched controls (single-image RL, joint-input RL, shuffled F2 feedback), and include a verifier reliability study. They also demonstrate transfer to urban scenes with the same rubric, which supports their claim of framework reusability. The ablations are the paper's strongest feature: they show that removing Gravity or Flight hurts, that aligned relational feedback beats shuffled feedback, and that the gain over supervised initialization (4.56 points) is not just from more compute. Credit where due—this is serious engineering with thorough documentation.
The soft spots are real but not disqualifying. The main one is the stress-test concern: a single run per setting, with no variance estimate for the headline score of 69.12 or the +4.56 gain over M1. The paper itself says that one run is reported per setting and that differences below a few percent should be read as run-to-run variation (b615), but that sentence is about wall-time cost, not evaluation scores. Given the stochastic sources (response sampling at temperature 0.6, random one-to-one matching, F2 repeat-change rate of 29.17%), the reader cannot assess whether the central numerical claim is robust. The verifier reliability is moderate—83.33% agreement within one point for F2 on 24 units—and the authors acknowledge that 24 units support an indication rather than a precise estimate (b475). The training data (500 pairs) and evaluation sets are small, and the method's dependence on a single fixed judge (DeepSeek-V4.1-Flash) means the results may not transfer to another verifier. These are all acknowledged, which is to the authors' credit, but they do limit confidence.
This paper is for researchers working on RL-based fine-tuning of VLMs for specialist domains, particularly remote sensing. It will be cited in that subfield. The method is clearly described and the ablations are informative. The main unresolved issue—single-run variance—is a standard weakness that peer review can address by asking for multi-seed experiments or a bootstrapped confidence interval. It does not undermine the paper's overall structure or claims.
Desk rejection would be too harsh. Send it to review. The paper deserves a serious referee who can ask for a variance estimate and push the authors to bound the run-to-run uncertainty.
The strongest objection
The paper reports a single run of each experimental setting (Section G: 'One run is reported per setting'). RTPO training involves multiple stochastic sources: (1) random response sampling at temperature 0.6, (2) random one-to-one matching of responses into candidate pairs, (3) verifier score variability as measured by repeat-change rates (F2: 29.17%, Table S4). Without multiple seeds, it is unknown whether the observed 4.56-point improvement over M1 is robust to these choices. The 100-unit training trace (Table S5) shows that 57.9% of pairs with one full-credit response receive a negative shared advantage, indicating that pair composition strongly affects credit assignment; a different random match could shift the advantage distribution and the resulting policy. The verifier's repeat-change rate of 29.17% for F2 means nearly a third of identical inputs yield different scores, adding noise to the paired-adaptation term. The paper acknowledges this variability (b472) but does not bound its impact on final evaluation scores. While ablations show consistent rank ordering (full RTPO > ablations), the magnitude of the gain over M1—a central claim—is not accompanied by any variance estimate. This is the most load-bearing concern because the headline numeric claim (69.12 overall, +4.56 over M1) could shift with run-to-run variation.
The reader identified verifier reliability as the weakest assumption. My concern extends to the combined effect of verifier noise, random response sampling, and random matching on the stability of the reported numeric gain. The paper's own reliability study (Table S4) shows non-trivial repeat-change rates, and Section G confirms single-run reporting. However, the extensive ablations, matched controls, and consistent patterns across settings provide converging evidence that the method improves over baselines in a reliable direction. The precise size of the improvement remains uncertain, but this does not invalidate the core contribution of the framework. Therefore I recommend the reader's ACCEPT verdict stand unchanged, though the authors should be expected to add multiple-run statistics in future work to strengthen the quantitative claim.
The referee's report
If the reported gains are robust, G&F represents a well-motivated and empirically grounded contribution to fine-grained visual reasoning under limited supervision. The separation of visual-fidelity verification (Gravity) from conditional-reasoning verification (Flight), combined with the paired-scene training design (RTPO), is novel and clearly explained. The paper provides extensive ablations (task-only, w/o Gravity, w/o Flight, single-image RL, paired RL w/o F2, shuffled F2, joint-input RL) that show a consistent rank ordering in favor of the full method. The urban specialization experiment demonstrates transfer to a second domain with unchanged rubric dimensions and weights, strengthening the claim of framework generality. The open release of code, data, and evaluation protocols supports reproducibility. The verifier reliability study (Table S4), while indicating moderate agreement for the most complex dimension (F2: 83.33% within one point, 29.17% repeat-change rate), is transparently reported and the limitations are acknowledged.
Pith's reading of the paper (a machine's summary, not the authors')
The paper argues that vision-language models reasoning over complex remote sensing scenes often miss fine-grained visual details and draw conclusions beyond what the evidence supports. To address this with limited supervision, the authors propose Gravity and Flight (G&F), a two-stage post-training framework that separates visual verification (Gravity) from reasoning verification (Flight) and couples policy updates across scene-matched image pairs. Evidence-of-Thought supervised fine-tuning teaches the model to describe visible details, analyse task-relevant evidence, and draw supported conclusions. Relational Transfer Policy Optimization then refines the policy on matched pairs where the same question applies to two images with fine-grained visual differences, using Gravity to check visual fidelity and coverage and Flight to evaluate evidence use and cross-scene judgment consistency. On six harbor evaluation sets, G&F achieves a mean score of 69.12, exceeding the backbone by 9.76 points and supervised initialization by 4.56 points, and matched controls show gains over single-image and joint-input reinforcement learning. A separate urban specialization also yields improvements, suggesting the framework can transfer to other scene domains.
What carries the argument. The key machinery is Relational Transfer Policy Optimization (RTPO), which operates on scene-matched image pairs. For each pair, the model generates responses independently for both images. A rubric-based verifier (DeepSeek-V4.1-Flash) assigns separate scores: Gravity (visual fidelity and description coverage) checked against the image and an Evidence Card; Flight (evidence use and paired adaptation) assessed conditional on the generated descriptions, using the public task and comparison metadata. These scores combine with task correctness into a pair reward. Rewards are group-normalized within each training unit to produce a shared pair-level advantage that updates the policy for both responses, coupling the learning across the two scenes.
Core claim. The central claim is that decoupling visual fidelity verification from conditional reasoning verification, and using a shared pair-level advantage derived from both, allows a vision-language model to learn fine-grained, evidence-grounded reasoning in remote sensing scenes with limited supervision (1,008 demonstrations and 500 image pairs). The discovery is that this separation—rather than a single end-to-end reward—provides more targeted training signals: Gravity catches visual fabrications and omissions, while Flight assesses whether the described evidence supports the answer, even when a premise is false. Pairing images with fine-grained differences under a shared task and coupling policy updates through the shared advantage forces the model to attend to task-relevant visual changes, leading to denser, better-supported reasoning in shorter responses.
The premise it rests on. The verifier (DeepSeek-V4.1-Flash) can reliably score visual fidelity, coverage, evidence use, and paired adaptation according to the rubrics, so that the policy update receives accurate reward signals; the paper acknowledges that judging errors may be correlated and provides a reliability study showing agreement within one point of human ratings on 83.33–100% of items across dimensions, but with limited sample size (24 units).
If the paper is right (Pith's inferences, not the authors')
If the claim holds, remotely-sensed scene understanding can be specialized with far fewer annotated examples than typical instruction-tuning requires, because the paired design extracts more signal per example.
The separation of visual and reasoning verification offers a general template for improving accuracy and faithfulness in vision-language tasks beyond remote sensing, such as medical imaging or autonomous driving scene analysis.
Because the shared advantage couples updates from two responses, the model learns to detect task-relevant visual differences even when the scenes are visually distinct but semantically similar, likely sharpening fine-grained discriminative ability.
The urban specialization experiment (using the same rubric dimensions and reward weights) suggests the framework may transfer across scene types without task-specific engineering, reducing the cost of deploying VLMs in new domains.
The verification design effectively acts as a learned reward model that is grounded in explicit rubrics, which may make the training more interpretable and less susceptible to reward hacking than a black-box scalar reward.
Paired training with a shared advantage implicitly encourages the model to compare scenes at the level of task-relevant evidence, a capability that could be directly tested on relational reasoning benchmarks outside the training domain.
The reliance on a single fixed verifier (DeepSeek-V4.1-Flash) raises a question of whether the gains would hold with other judges; a sensitivity study across verifiers would help establish robustness.
Because the advantage is normalized within each pair set, the method may be sensitive to the composition of the training units—units with many invariant or easily distinguished pairs could produce weaker training signals, potentially affecting convergence.
Check this review
Pith stored no record of this review's passes and took no fingerprint of it. Reviews made by code from 13 Nov 2026 on carry both.
How often reviews like this one are right
Reviews like this one: written by deepseek-v4-flash, where what was read was not recorded.
How often objections from reviews like this one are right: not yet measured
Reviews by people
This review was written by a machine. No person has reviewed this paper on Pith yet. A scientist listed on Pith can review it for a fee, and their review is published here whatever it concludes.
The paper
Abstract
Vision-language models are increasingly capable, yet their reasoning in complex remote sensing scenes can remain coarse-grained and insufficiently grounded in task-relevant visual evidence. We propose Gravity and Flight (G&F), a framework for scene specialization with limited supervision that separates visual and reasoning verification while coupling policy updates. It combines Evidence-of-Thought supervised fine-tuning (EoT-SFT) and Relational Transfer Policy Optimization (RTPO). EoT-SFT teaches fine-grained observation and faithful reasoning through description, analysis, and supported conclusions. RTPO independently samples responses to scene-matched image pairs with fine-grained visual differences under a shared task. Its rubric-based verification combines Gravity, which checks visual fidelity and description coverage, and Flight, which takes descriptions as premises to assess evidence use and relations between paired judgments. Their scores and task correctness yield a shared pair-level advantage that couples policy updates from both responses. Using 1,008 demonstrations and 500 training pairs for harbor specialization, G&F achieves a mean score of 69.12 across six evaluation sets, exceeding the backbone and supervised initialization by 9.76 and 4.56 points. Matched controls show gains over single-image and joint-input RL, and removing either verification view reduces performance. Applying G&F to urban specialization also improves on the base model and supervised initialization across three evaluation subsets. Project details and updates are available at https://github.com/Z0ngL1ng/Gravity-and-Flight.
The paper's own sentences, arranged, kept beside the review and separate from it. Each label is drawn from a closed vocabulary; the number beside it is the probability the classifier gave that label. No sentence in this record was written by a model.
abstract compiled · 2026-11-13
The authors' main claim, as the classifier picked it out 0.89
“Using 1,008 demonstrations and 500 training pairs for harbor specialization, G&F achieves a mean score of 69.12 across six evaluation sets, exceeding the backbone and supervised initialization by 9.76 and 4.56 points.”
How the main claim is stated: asserted 1.00
What the abstract itself states, as probabilities: claims first 0.04 · states limitation 0.17 · human participants 0.06 · controlled comparison 0.98 · randomized assignment 0.02 · states quantitative result 0.99 · compares with prior results 0.98 · mentions code or data availability 0.89
The abstract, sentence by sentence
This table labels the abstract only (arXiv listing abstract v1). The number is the probability the classifier gave the label.
| Sentence | Role | The authors' own work | States the main conclusion | Strength |
|---|---|---|---|---|
| Vision-language models are increasingly capable, yet their reasoning in complex remote sensing scenes can remain coarse-grained and insufficiently grounded in task-relevant visual evidence. | background 0.60 | 0.08 | 0.04 | asserted 0.72 |
| We propose Gravity and Flight (G&F), a framework for scene specialization with limited supervision that separates visual and reasoning verification while coupling policy updates. | contribution 0.95 | 0.86 | 0.10 | asserted 0.99 |
| It combines Evidence-of-Thought supervised fine-tuning (EoT-SFT) and Relational Transfer Policy Optimization (RTPO). | method 0.93 | 0.66 | 0.04 | asserted 1.00 |
| EoT-SFT teaches fine-grained observation and faithful reasoning through description, analysis, and supported conclusions. | method 0.98 | 0.55 | 0.04 | asserted 1.00 |
| RTPO independently samples responses to scene-matched image pairs with fine-grained visual differences under a shared task. | method 0.99 | 0.86 | 0.03 | asserted 0.99 |
| Its rubric-based verification combines Gravity, which checks visual fidelity and description coverage, and Flight, which takes descriptions as premises to assess evidence use and relations between paired judgments. | method 0.99 | 0.79 | 0.06 | asserted 1.00 |
| Their scores and task correctness yield a shared pair-level advantage that couples policy updates from both responses. | method 1.00 | 0.91 | 0.06 | asserted 1.00 |
| Using 1,008 demonstrations and 500 training pairs for harbor specialization, G&F achieves a mean score of 69.12 across six evaluation sets, exceeding the backbone and supervised initialization by 9.76 and 4.56 points. | result 1.00 | 0.97 | 0.42 | asserted 1.00 |
| Matched controls show gains over single-image and joint-input RL, and removing either verification view reduces performance. | result 0.98 | 0.95 | 0.11 | asserted 1.00 |
| Applying G&F to urban specialization also improves on the base model and supervised initialization across three evaluation subsets. | result 1.00 | 0.95 | 0.13 | asserted 1.00 |
| Project details and updates are available at https://github.com/Z0ngL1ng/Gravity-and-Flight. | future work 0.78 | 0.14 | 0.01 | asserted 0.89 |
What kind of paper the classifier reports this is: method or tool 1.00 · evidence from observation or experiment 0.89
What was examined, and what was not
Sources read
- arXiv listing abstract v1
Not examined
- full text
- figures and tables
- references and citing sentences
- relations to other papers
Levels run
- P abstract
Not run
- S sections
- T body sentences
- C citations
- R cross-paper relations
How the labels were made
Compiled by classifier-1.13.0 via TypeSafe API, question bank abstract-p1-20260923, 2026-11-13: 1 request, 8,200 input tokens.
Every quoted sentence on this record is the authors'. Every label is from Pith's closed vocabulary, with the probability shown beside it. This record as JSON
Figures (4)
Figures are the authors', reproduced from the paper.
Challenges and discussion
Disagree with the review, or know something it missed? Leave a comment, or challenge this review formally: a challenge becomes a permanent entry in this paper's record, and its decision is published here.
Come back by email
A short list of the week's newly reviewed papers. Nothing is sent until you confirm from your inbox, and every email has a link that stops it. No account.
No comments yet.
Sign in to comment