High-resolution phenotyping at the level of individual leaves offers fine-grained insights into plant development and stress responses. However, the full potential of accurate leaf tracking over time remains largely unexplored due to the absence of robust tracking methods, particularly for structurally complex crops such as canola. Existing plant-specific tracking methods are typically limited to small-scale species or rely on constrained imaging conditions. In contrast, generic multi-object tracking (MOT) methods are not designed for dynamic biological scenes. Progress in the development of accurate leaf tracking models has also been hindered by a lack of large-scale datasets captured under realistic conditions. In this work, we introduce CanolaTrack, a new benchmark dataset comprising 5704 RGB images with 31,840 annotated leaf instances collected from 184 canola plants during their early growth stages. To enable accurate leaf tracking over time, we introduce LeafTrackNet, an efficient framework that combines a YOLOv10-based leaf detector with a MobileNetV3-based embedding network. During inference, leaf identities are maintained over time through an embedding-based memory association strategy. When trained directly on each target dataset without prior CanolaTrack fine-tuning, LeafTrackNet achieves HOTA scores of 88.03, 87.33, and 74.20 on CanolaTrack, KOMATSUNA, and MSU-PID, respectively, outperforming the corresponding second-best methods by 8.35, 4.94, and 1.62 HOTA points. All results are reported as mean ± standard deviation over three independent runs, using fixed inference hyperparameters across datasets without per-dataset tuning. This work provides a new benchmark for leaf-level tracking under realistic conditions and introduces CanolaTrack, which, to the best of our knowledge, is the largest leaf-tracking dataset for agricultural crops. Our code and dataset are publicly available at GitHub.
This study investigated whether preschool children act on the understanding that identical-looking objects can no longer be re-identified once their positions change unnoticed, using an everyday task assessing the ability to avoid misidentification and comparing their performance with that of adults. Although children aged 4 to 7 years (N = 66) reliably and repeatedly identified which of five identical-looking nuggets was a Playmobil figure’s “favorite”, the majority (56 %) nevertheless failed to separate it when tidying up the five nuggets into two opaque bags. In contrast, all adults (N = 66) did so. The fact that the majority of children failed to separate the verbally marked nugget, despite knowing “which one it was”, indicates that they struggle to anticipate that the identity of very similar objects can be maintained only as long as identity-relevant spatial information is available, pointing to a still-developing concept of object identity. Older preschoolers tended to be more successful in separating the figure’s favorite nugget than younger ones, suggesting a developmental progression in how spatial identity criteria for objects are conceptualized. In a condition in which the figure’s favorite nugget was perceptually distinct (different color), 64 % of another group of children (N = 66) packed it separately, even though this was not necessary, indicating that their behavior was strongly driven by perceptual salience. Taken together, these findings suggest that many preschool children still struggle to use spatial criteria of identity to guide their behavior, even in a context that minimizes memory, attentional, and linguistic demands.
The ultraviolet (UV) spectral domain occupies a unique position in stellar astrophysics, serving as the bridge between the thermal continuum of photospheres and the high-energy, non-thermal processes of stellar coronae and winds. This article provides a review of stellar physics in the UV, addressing both the theoretical framework and observational applications across the Hertzsprung-Russell diagram. We explicitly structure our discussion around key scientific questions, demonstrating that accurate spectral synthesis in this regime demands Non-LTE radiative transfer codes, which in turn rely on precise atomic collision and recombination rates. We highlight how a critical scarcity of modern laboratory astrophysics data limits these models, particularly for complex ions. Moving to observational diagnostics, we review how UV spectroscopy constrains diffusion and radiatively driven winds in hot subluminous stars, and traces shock dynamics and abundance patterns in Planetary Nebulae and Supernova Remnants. In the context of star clusters, we illustrate how UV sensitivity to light-element variations (C, N, O) allows us to disentangle multiple stellar populations that appear degenerate in optical bands. We conclude that future progress depends on facilities capable of high-resolution spectroscopy, time-domain monitoring, and polarimetry to recover these diagnostic tracers and resolve the physics of stellar feedback.
Context. Hot subdwarf B (sdB) and O (sdO) type stars are evolved helium-burning objects that lost their hydrogen envelope before the helium flash when their progenitors were close to the tip of the red giant branch (RGB). They populate the extreme horizontal branch (EHB) in the Hertzsprung-Russell diagram (HRD). The mass distribution of canonical hot subdwarfs is expected to peak at the core mass required for helium ignition under degenerate conditions in the 0.45-0.5 M⊙ range. However, non-degenerate helium ignition from intermediate-mass progenitors and non-canonical pathways, such as the merger of helium white dwarfs and delayed helium flashes, are also expected to contribute to the hot subdwarf population. Aims. Using high-quality, homogeneous spectra of 335 hot subluminous star candidates from the Arizona-Montréal Spectroscopic Survey, we aim to improve our understanding of the atmospheric and stellar properties of hot subdwarf stars. Our focus is on the mass distribution of the different types of hot subdwarfs and their connections to the various formation scenarios. Methods. We used large grids of model atmospheres to fit the observed spectra and derived their atmospheric parameters: effective temperature (Teff), surface gravity, and helium abundance. The model grids were further utilized to fit the spectral energy distribution of each star and the Gaia parallax was used to compute the stellar parameters radius, luminosity, and mass. Results. Our spectroscopic sample mostly consists of H-rich sdBs and sdOs, but also contains 41 He-rich sdOs. Additionally, the sample includes 11 intermediate-helium stars and 19 horizontal branch objects with Teff > 14 kK. We detected the presence of helium stratification in six sdB stars with Teff around 30 kK, making them good candidates for also showing 3He enrichment in their atmospheres. Our sdB distribution along the EHB shows a gap near 33 kK, visible in both the Kiel (logg-Teff) diagram and HRD, corroborating previous observations and predictions. The mass distributions of H-rich sdBs and sdOs are similar and centered around 0.47 M⊙, consistent with the canonical formation scenario of helium ignition under degenerate conditions. Among the H-rich hot subdwarfs, we found no difference between the mass distributions of close binaries and apparently single stars. The He-sdOs have a significantly wider mass distribution than their H-rich counterparts, with an average mass of about 0.78 M⊙. In the HRD, the He-sdOs lie on the theoretical helium main sequence for masses between 0.6 and 1 M⊙. This strongly favors a merger origin for these He-rich objects. We identified a small number of candidate low-mass (<0.45 M⊙) sdBs located below the EHB that might have originated from more massive progenitors. These low-mass sdBs preferentially show low helium abundances. Finally, we identified more than 80 pulsating stars in our sample and found that they fall into well-defined p- and g-mode instability regions.
Context. Dusty discs detected around main-sequence stars are thought to be signs of planetesimal belts in which the dust distribution is shaped by collisional and dynamical processes, including interactions with gas if present. The debris disc around the young A-type star HD 131835 is composed of two dust rings at similar to 65 au and similar to 100 au, a third unconstrained innermost component, and a gaseous component centred at similar to 65 au. New ALMA observations show that the inner of the two dust rings is brighter than the outer one, in contrast with previous observations in scattered light. Aims. We explore two scenarios that could explain these observations: the two dust rings might represent distinct planetesimal belts with different collisional properties, or only the inner ring might contain planetesimals while the outer ring consists entirely of dust that has migrated outwards due to gas drag. Methods. To explore the first scenario, we employed a state-of-the-art collisional evolution code. To test the second scenario, we used a simple dynamical model of dust grain evolution in an optically thin gaseous disc. In each case we identified the parameters of the planetesimal and the gaseous disc that best reproduce the observational constraints. Results. Collisional models of two planetesimal belts cannot fully reproduce the observations by only varying their dynamical excitation, and matching the data through a different material strength requires an extreme difference in dust composition. The gas-driven scenario can reproduce the location of the outer ring and the brightness ratio of the two rings from scattered light observations, but the resulting outer ring is too faint overall in both scattered light and sub-millimetre emission. Conclusions. The dust rings in HD 131835 could be produced from two planetesimal belts, although how these belts would attain the required extremely different properties needs to be explained. The dust-gas interaction is a plausible alternative explanation and deserves further study using a more comprehensive model.
