arXiv subjects

Joseph Nelson

Publications and source records attributed to Joseph Nelson.

4 recordsLinked to original sources

arXiv2025

Roboflow100-VL: A Multi-Domain Object Detection Benchmark for Vision-Language Models

Vision-language models (VLMs) trained on internet-scale data achieve remarkable zero-shot detection performance on common objects like car, truck, and pedestrian. However, state-of-the-art models still struggle to generalize to out-of-distribution classes, tasks and imaging modalities not typically found in their pre-training. Rather than simply re-training VLMs on more visual data, we argue that one should align VLMs to new concepts with annotation instructions containing a few visual examples and rich textual descriptions. To this end, we introduce Roboflow100-VL, a large-scale collection of 100 multi-modal object detection datasets with diverse concepts not commonly found in VLM pre-training. We evaluate state-of-the-art models on our benchmark in zero-shot, few-shot, semi-supervised, and fully-supervised settings, allowing for comparison across data regimes. Notably, we find that VLMs like GroundingDINO and Qwen2.5-VL achieve less than 2% zero-shot accuracy on challenging medical imaging datasets within Roboflow100-VL, demonstrating the need for few-shot concept alignment. Lastly, we discuss our recent CVPR 2025 Foundational FSOD competition and share insights from the community. Notably, the winning team significantly outperforms our baseline by 17 mAP! Our code and dataset are available at https://github.com/roboflow/rf100-vl and https://universe.roboflow.com/rf100-vl/.

cs.CV↗

arXiv2017

A novel recommendation system to match college events and groups to students

With the recent increase in data online, discovering meaningful opportunities can be time-consuming and complicated for many individuals. To overcome this data overload challenge, we present a novel text-content-based recommender system as a valuable tool to predict user interests. To that end, we develop a specific procedure to create user models and item feature-vectors, where items are described in free text. The user model is generated by soliciting from a user a few keywords and expanding those keywords into a list of weighted near-synonyms. The item feature-vectors are generated from the textual descriptions of the items, using modified tf-idf values of the users' keywords and their near-synonyms. Once the users are modeled and the items are abstracted into feature vectors, the system returns the maximum-similarity items as recommendations to that user. Our experimental evaluation shows that our method of creating the user models and item feature-vectors resulted in higher precision and accuracy in comparison to well-known feature-vector-generating methods like Glove and Word2Vec. It also shows that stemming and the use of a modified version of tf-idf increase the accuracy and precision by 2% and 3%, respectively, compared to non-stemming and the standard tf-idf definition. Moreover, the evaluation results show that updating the user model from usage histories improves the precision and accuracy of the system. This recommender system has been developed as part of the Agnes application, which runs on iOS and Android platforms and is accessible through the Agnes website.

cs.IR↗

arXiv2015

Quantum Hydrogen-Bond Symmetrization and High-Temperature Superconductivity in Hydrogen Sulfide

Hydrogen compounds are peculiar as the quantum nature of the proton can crucially affect their structural and physical properties. A remarkable example are the high-pressure phases of H$_2$O, where quantum proton fluctuations favor the symmetrization of the H bond and lower by 30 GPa the boundary between the asymmetric structure and the symmetric one. Here we show that an analogous quantum symmetrization occurs in the recently discovered sulfur hydride superconductor with the record superconducting critical temperature $T_c=203$ K at 155 GPa. In this system, according to classical theory, superconductivity occurs via formation of a structure of stoichiometry H$_3$S with S atoms arranged on a body-centered-cubic (bcc) lattice. For $P \gtrsim 175$ GPa, the H atoms are predicted to sit midway between two S atoms, in a structure with $Im\bar3m$ symmetry. At lower pressures the H atoms move to an off-center position forming a short H$-$S covalent bond and a longer H$\cdots$S hydrogen bond, in a structure with $R3m$ symmetry. X-ray diffraction experiments confirmed the H$_3$S stoichiometry and the S lattice sites, but were unable to discriminate between the two phases. Our present ab initio density-functional theory (DFT) calculations show that the quantum nuclear motion lowers the symmetrization pressure by 72 GPa. Consequently, we predict that the $Im\bar3m$ phase is stable over the whole pressure range within which a high $T_c$ was measured. The observed pressure-dependence of $T_c$ is closely reproduced in our calculations for the $Im\bar3m$ phase, but not for the $R3m$ phase. Thus, the quantum nature of the proton completely rules the superconducting phase diagram of H$_3$S.

cond-mat.supr-con↗

arXiv2015

Hydrogen sulphide at high pressure: a strongly-anharmonic phonon-mediated superconductor

We use first principles calculations to study structural, vibrational and superconducting properties of H$_2$S at pressures $P\ge 200$ GPa. The inclusion of zero point energy leads to two different possible dissociations of H$_2$S, namely 3H$_2$S $\to$ 2H$_3$S + S and 5H$_2$S $\to$ 3H$_3$S + HS$_2$, where both H$_3$S and HS$_2$ are metallic. For H$_3$S, we perform non-perturbative calculations of anharmonic effects within the self-consistent harmonic approximation and show that the harmonic approximation strongly overestimates the electron-phonon interaction ($λ\approx 2.64$ at 200 GPa) and T$_c$. Anharmonicity hardens HS bond-stretching modes and softens H--S bond-bending modes. As a result, the electron-phonon coupling is suppressed by $30\%$ ($λ\approx 1.84$ at 200 GPa). Moreover, while at the harmonic level T$_c$ decreases with increasing pressure, the inclusion of anharmonicity leads to a T$_c$ that is almost independent of pressure. High pressure hydrogen sulfide is a strongly anharmonic superconductor.

cond-mat.supr-con↗