REVIEW 4 major objections 2 minor 1 cited by
IMUCoCo: Enabling Flexible On-Body IMU Placement for Human Pose Estimation and Activity Recognition
T0 review · 4 major / 2 minor · reviewed 2026-08-06 · deepseek-v4-flash
Pith's one-line read IMUCoCo claims that pairing each IMU signal with its body-surface coordinate and mapping these pairs into a unified feature space lets pose estimation and activity recognition work across a wide range of sensor placements, including…
desk verdict The full text under this arXiv ID is a cosmology paper, so the only thing on the table is an abstract that promises a useful idea but leaves the key questions unanswered. read the letter →
The pith
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
The reading
What carries the argument
The load-bearing mechanism is the continuous-coordinate conditioning step: each IMU's inertial signal is combined with the IMU's body-surface coordinate and encoded into a unified feature space, so downstream models see placement-aware features rather than signals tied to named body sites. Because the conditioning variable is continuous, the same encoder can in principle accept any number of sensors at any location on the body surface. The paper names this framework IMU over Continuous Coordinates (IMUCoCo).
What would settle it
Train an IMUCoCo model on wrist, chest, and ankle placements, then evaluate it with a single IMU placed on the crown of the head or the sole of the foot; if pose error is no better than a coordinate-blind baseline, the claimed flexibility is not supported.
Extended reading notes
Core claim
The central claim is that sensor location can be treated as a continuous input rather than a discrete categorical slot. By conditioning each IMU's signal on its body-surface coordinate and projecting the result into a unified feature space, IMUCoCo aims to make the learned representation independent of any specific placement configuration. The paper reports that this representation supports accurate pose estimation across a wide range of typical and atypical placements, works with a variable number of sensors, and also enables use-case-dependent placement suggestions.
Load-bearing premise
The framework assumes that each IMU's body-surface coordinate is known or inferable at inference time and that the coordinate-conditioned mapping learned from training placements generalizes to body locations and sensor counts never seen in training.
Editorial extensions
If this is right
- A user could move an IMU-equipped device from wrist to ankle, pocket, or elsewhere between sessions or contexts without retraining, provided the new coordinate is supplied.
- Because the feature space is unified, the same downstream pose and activity models can accept different numbers of IMUs, so adding or removing a sensor does not require a new model.
- The placement-suggestion capability could tell users where to put sensors to maximize accuracy for a specific activity or use case.
- If the mapping generalizes smoothly, sparse training placements could support many untrained body locations rather than only the handful of predefined spots used today.
- Both pose estimation and activity recognition can be plugged into the same IMUCoCo feature space, so one sensing framework serves both tasks.
Reading between the lines
- The supplied full text is a different manuscript about primordial black hole clustering, so the abstract and metadata are the only evidence for IMUCoCo's empirical claims in this record.
- If the coordinate-to-feature mapping is continuous over the body surface, features for unmeasured locations could be interpolated, in effect creating virtual sensors where no device is worn; the paper does not claim this directly.
- A practical deployment likely needs a way to obtain each IMU's body-surface coordinate automatically, for example by self-localization from the IMU signals, because manual measurement would undermine the flexibility the framework promises.
- Training on heterogeneous placements from many users could enlarge the available motion data, but cross-user use may require normalizing coordinates across different body shapes.
Signed reviews
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The submission presents IMUCoCo, a framework intended to map IMU signals from an arbitrary, variable number of body-surface sensors into a unified feature space conditioned on spatial coordinates, with downstream pose estimation and activity recognition. The abstract claims that the framework supports accurate pose estimation across typical and atypical sensor placements, enables users to change device locations by context, and suggests placements per use case. However, the supplied full text is not the IMUCoCo paper but an unrelated cosmology manuscript on primordial black holes in excursion set theory. Consequently, the submission contains no method description, no training or inference details, no evaluation protocol, no baselines, and no quantitative results for the claimed system. Every central claim in the abstract is therefore unverifiable from the submitted material.
Significance. If the claimed framework were actually implemented and validated, it would address a genuine and growing usability bottleneck in IMU-based motion sensing: freeing users from a small set of fixed, training-dictated sensor placements and allowing flexible, context-dependent placement. The problem statement is well motivated, and the idea of a continuous coordinate-conditioned feature space is a plausible and potentially valuable design direction. That said, the submission as received provides no evidence for the central claims and no artifact that could be checked: there are no machine-checked proofs, no reproducible code, no datasets, and no empirical results. The significance can only be judged as prospective, not demonstrated.
major comments (4)
- [Full Text (submitted manuscript)] The full text under the submission is arXiv:2508.01896, 'Clustering of Primordial Black Holes in Excursion Set Theory', which is entirely unrelated to the IMUCoCo abstract. The manuscript therefore contains none of the claimed framework: no network architecture, no training objective, no coordinate encoding scheme, no downstream model, no evaluation protocol, and no results. This is a load-bearing failure because the central claim of flexible, accurate IMU-based pose estimation cannot be assessed in any way. This issue cannot be fixed by local revision; the actual IMUCoCo manuscript would have to be supplied in its entirety.
- [Abstract] The sentence 'Our evaluations demonstrate that IMUCoCo supports accurate pose estimation in a wide range of typical and atypical sensor placements' is made without any quantitative support: no dataset is named, no baseline is compared, no error metric is reported, no ablation is described, and no error analysis is given. For a machine-learning systems paper, an empirical claim of state-of-the-art flexibility requires at least summary statistics and a clear held-out evaluation protocol; none appear in the submission.
- [Abstract] The framework's premise is that each IMU's spatial coordinate on the body surface is available at inference time, yet the abstract never states how these coordinates are obtained: by manual annotation, by device self-localization, by a learned estimator, or by some other mechanism. If a user must measure or enter coordinates at every placement, the promised flexibility is weakened by an unstated user burden; if coordinates come from another model, errors in that estimate propagate into the unified feature space. This condition is load-bearing for the central flexibility claim and is left unspecified.
- [Abstract] The claim that the model generalizes to 'a wide range of typical and atypical sensor placements' and to 'a variable number of IMUs' is not tied to any explicit training distribution or evaluation split. The abstract does not establish that the coordinate-conditioned mapping is tested on held-out placements and sensor counts rather than on placements that shaped the model during training, leaving the core flexibility claim unsubstantiated.
minor comments (2)
- [General] The title, abstract, and full text describe entirely different papers; even a reader who only skims the submission will be confused about the subject matter.
- [General] The submission contains no references, figures, or equations for the IMUCoCo work, making it impossible to follow or verify any part of the proposed method.
Circularity Check
No circularity found: the submitted full text is a different manuscript, so there is no IMUCoCo derivation chain to evaluate.
full rationale
The submitted full text under arXiv:2508.01894 is not the IMUCoCo paper. Its opening lines read "Clustering of Primordial Black Holes in Excursion Set Theory / Hamed Kameli ... E-mail: hkameli@gmail.com, eerfani@perimeterinstitute.ca", and the abstract describes a cosmology result about PBH pair formation in Excursion Set Theory. Consequently, none of the IMUCoCo methodology, equations, training procedure, or evaluation protocol is present in the provided text. The IMUCoCo abstract alone contains no fitted parameters, no equations, and no self-citations, so there is no derivation chain whose output could be shown to reduce to its input by construction. Under the hard rule that circularity may only be claimed when a specific reduction can be quoted and exhibited, no such step exists here. The abstract's silence on how sensor coordinates are obtained and whether held-out placements were tested is a verifiability or completeness concern, not a circularity concern. The verdict is therefore a non-finding: score 0, no circular steps identified.
Assumptions & free parameters
assumptions (2)
- domain assumption IMU signals at a known body-surface coordinate contain enough information to infer full-body pose and activity from any placement.
- domain assumption The user's system can provide or infer accurate spatial coordinates for each IMU at inference time.
Cite this review
Pith. "Pith review of IMUCoCo: Enabling Flexible On-Body IMU Placement for Human Pose Estimation and Activity Recognition." pith.science (2026). https://pith.science/paper/4FT6Z4IF
@misc{pith2026250801894,
author = {Pith},
title = {Pith review of: IMUCoCo: Enabling Flexible On-Body IMU Placement for Human Pose Estimation and Activity Recognition},
year = {2026},
howpublished = {\url{https://pith.science/paper/4FT6Z4IF}},
note = {Machine review of arXiv:2508.01894}
}
read the original abstract
IMUs are regularly used to sense human motion, recognize activities, and estimate full-body pose. Users are typically required to place sensors in predefined locations that are often dictated by common wearable form factors and the machine learning model's training process. Consequently, despite the increasing number of everyday devices equipped with IMUs, the limited adaptability has seriously constrained the user experience to only using a few well-explored device placements (e.g., wrist and ears). In this paper, we rethink IMU-based motion sensing by acknowledging that signals can be captured from any point on the human body. We introduce IMU over Continuous Coordinates (IMUCoCo), a novel framework that maps signals from a variable number of IMUs placed on the body surface into a unified feature space based on their spatial coordinates. These features can be plugged into downstream models for pose estimation and activity recognition. Our evaluations demonstrate that IMUCoCo supports accurate pose estimation in a wide range of typical and atypical sensor placements. Overall, IMUCoCo supports significantly more flexible use of IMUs for motion sensing than the state-of-the-art, allowing users to place their sensors-laden devices according to their needs and preferences. The framework also supports the ability to change device locations depending on the context and suggests placement depending on the use case.
Forward citations
Cited by 1 Pith paper
-
BVQC: A Backdoor-style Watermarking Scheme for Variational Quantum Circuits
BVQC embeds a backdoor watermark into variational quantum circuits so base task loss is preserved in normal use but rises to a predefined level during watermark extraction, claiming improvements in recompilation robus...
Reference graph
Works this paper leans on
-
[1]
Clustering of Primordial Black Holes in Excursion Set Theory
Clustering of Primordial Black Holes in Excursion Set Theory Hamed Kameli a Encieh Erfani b aDepartment of Physics, Sharif University of Technology, Tehran 11155-9161, Iran bPerimeter Institute for Theoretical Physics, Waterloo, ON N2L 2Y5, Canada E-mail: hkameli@gmail.com, eerfani@perimeterinstitute.ca Abstract. We investigate the clustering of Primordia...
work page Pith review arXiv 2025
Reviewed August 6, 2026 · model on record in the stance chip above.
Discussion (0). Continue with ORCID to comment.