Mendeley Data Showcase

The dataset is composed of high fidelity images of a physical planetary analog environment that are annotated with foreground and background semantic labels and were created for training lightweight semantic segmentation models (e.g., SegFormer) for autonomous rover navigation and ingestion of the Semantic Digital Twin (SDT). To help close the gap between the simulation and reality, a physical analog "sandbox" test bed was built to simulate both Martian terrain and hazard terrain, with dynamic obstacles, different soil properties and rock structures. An edge-computing setup (Raspberry Pi 4B with a Sony IMX708 Camera Module 3) captured a continuous data stream, resulting in 500 images being taken as raw data. The out-of-focus and redundant frames were rejected and the final baseline consisted of 482 high data quality images. These images have been uploaded to the Roboflow platform where pixel-perfect, polygon-based semantic segmentation masks have been handcrafted for each image, for a specific terrain class and hazard class. - The most efficient way to split a dataset into three or more segments. The dataset was thoroughly split into 70% training, 15% validation and 15% test sets to prevent data leakage and provide a strong evaluation. Training Set (70%): ~337 images Validation Set (15%): ~72 images Testing Set (15%): ~72 images Data Augmentation: The data augmentation was applied only to the 70% Training Set, in order to improve the generalization performance of the neural network, while keeping the evaluation metrics as they are. These training frames were then augmented by a 3x factor by methods like random rotation, cropping and contrast changes. This augmented training data was then re-combined with the strictly un-augmented validation and testing sets, resulting in a total of ~1,155 images ready to be used in edge-AI training pipelines. Mixed_dataset_70_30: In addition to the primary custom dataset described above, this repository also includes a Mixed_dataset_70_30 archive. This represents a 70-30 domain-blended mixture, created by combining our ~1,155 custom augmented images with randomly sampled frames from the public S5Mars source domain dataset. This blended dataset is provided alongside the custom dataset to ensure 100% reproducibility of the domain-mixture experiments and results discussed in our associated research. Potential Applications: Lightweight training of Vision Transformers and CNNs for semantic segmentation. Planetary micro rover Hardware-in-the-loop testing. Testing of bandwidth optimization and telemetry throttling, and research for deep-space communications.

Replication package for "The Digital Divide in Firm Resilience: New Evidence from Global Crises", submitted to Economic Modelling.

Table S1. Sample locations, occurrence, and mineralogy in this study. Table S2 is the reference standard materials associated with analytical methods in this study. Table S3. Table S3. Major composition of apatite in the Maoniuping REE deposit. Table S4. Table S4. Trace element compositions of apatite from the Maoniuping REE deposit. Table S5. In-situ Rb-Sr isotopic compositions of apatite from the Maoniuping REE deposit. Table S6. In situ Sm-Nd isotopic compositions of apatite in the Maoniuping REE deposit. Table S7. In situ O isotopic composition of apatite from the Maoniuping deposit.

This dataset contains 256×256 pixel patches extracted from multispectral UAV orthomosaics acquired over barley (Hordeum vulgare) and rapeseed (Brassica napus) fields in Valdeavero (Madrid, Spain). It includes four subsets: Barley 1‑V1 (training, 726 patches), Barley 1‑V2 (temporal validation, 524 patches), Barley 2‑V2 (spatial validation), and Rapeseed‑V2 (cross‑crop transfer and fine‑tuning, 423 patches). Each patch has 8 spectral bands: R, G, B, NIR, RedEdge, NDVI, NDRE, and VARI. The dataset also includes text files defining the splits used for few‑shot fine‑tuning experiments (5%, 10%, and 15%).Agriculture

This dataset contains household survey data and Stata replication code for a study of the drivers and expected timing of farm succession and exit among smallholder upland farming households in Lantapan, Bukidnon, Philippines. The data were collected in 2024 through a tablet-based survey (KoboCollect) of 314 farming households, drawn by stratified random sampling (strata by elevation, Neyman allocation) from the 5,177 landowner farmers on the Master List of Farmers of the Municipal Agriculture Office. Respondents reported the likelihood of intra-family farm succession on a five-category scale, the expected number of years until succession or exit, and detailed information on household composition, the household head’s and spouse’s characteristics, children’s farm involvement and migration plans, farm income and assets, land, production and marketing practices, shocks experienced, and intra-household decision-making measured through a Women’s Empowerment in Agriculture Index (WEAI) adapted from Alkire et al. (2013) with stricter adequacy cut-offs for the Philippine context. The deposit has two parts. The Excel workbook (Data for Analysis.xlsx) holds the anonymized analysis dataset: one row per household, identified only by an anonymous code and barangay, in four sheets (Probit for the succession-likelihood models; AFT_All for the competing-risks timing models; AFT_SUC and AFT_Exit as single-pathway reference sheets). The two Stata do-files reproduce every table and figure of the associated article: probit do file estimate the drivers of expected succession on the full sample (n = 314) and on the headline sample excluding undecided households (n = 266), with descriptives, diagnostics, alternative estimators, and multiple-imputation robustness; the accelerated-failure-time (AFT) file constructs the survival data, selects the failure-time distribution by AIC/BIC (Weibull for both pathways), and estimates the timing of succession and exit as competing risks with barangay fixed effects and robust standard errors. All monetary variables are in Philippine pesos (PHP). Informed consent was obtained from all participants and no direct identifiers are included.

This dataset contains the results of an indicator-based assessment of the accessibility and usability of Energy Performance Certificate (EPC) information systems in three Spanish Autonomous Communities: Catalonia, Galicia, and the Basque Country. The study examines whether the decentralised management of EPC registers produces regional differences despite a common national and European regulatory framework. The framework comprises two analytical dimensions, eleven assessment criteria, and forty-five operational indicators covering access, data availability, retrievability, visualisation, openness, interoperability, dataset comprehensiveness, harmonisation, and metadata quality. Data were collected between April and December 2025 through direct observation of official EPC registers, institutional websites, open-data portals, geoportals, visualisation platforms, and downloadable datasets. Each region was independently assessed by two researchers using a common protocol, and discrepancies were resolved by consensus. The workbook includes indicator-level scores, aggregated results by criterion and dimension, weighted overall scores, comparative tables, and graphs. Scores range from 0 to 100, with higher values indicating better accessibility or usability. The overall weighted scores are 83.27 for the Basque Country, 77.89 for Catalonia, and 64.62 for Galicia. The dataset can be used to reproduce the comparative results reported in the associated article, identify strengths and weaknesses in regional EPC information systems, and support future applications of the framework in other territorial or national contexts.

This dataset contains data from the experiment reported in the paper "Dear AI, Which Charging Station Should I Pick? The Role of Uncertainty and Perceived Functionality Trust in Charging Station Selection Through AI Recommendations." In the experiment, participants faced a charging task where they had to select between two charging stations in 36 scenarios where expected travel time and range (i.e., maximum minus minimum travel time) were systematically manipulated. Then, participants answered some questions related to perceived functionality trust in an app that assisted them while making the decisions, as well as perceived functionality trust in navigation apps in general and in smart technology. Additionally, participants answered questions related to numeracy, risk propensity, charging behaviors, and demographic information.

Hyperspectral reflectance images for 16 virgin plastic objects that were captured using the Specim FX10 (400 – 1000 nm, 224 bands) and Specim FX17 (900 – 1700 nm, 224 bands) sensors. Samples were scanned in continuous mode on a motorised stage under halogen illumination. Raw data were calibrated to reflectance using a Spectralon white calibration standard and dark reference (lens covered). Raw data, white and black references, and resultant reflectance are provided for each sensor. The calibrated FX17 is spatially aligned (RESAMP_CAL_Plastic_FX17_0005) with the FX10 data to identify the Regions Of Interest (ROIs) for later pixel extraction and classification. Classes shape contains ROIs for each class (the polygons .shp, .shx, and .dbf). Objects: 1. Fruit net (suspected HDPE) 2. Transparent plastic bag (suspected HDPE) 3. White plastic bag (suspected LDPE) 4. White cereals plastic bag (suspected HDPE) 5. Toothpaste (ground truth HDPE) 6. Toothpaste cap (suspected PP) 7. Sanex shampoo cap (suspected HDPE) 8. Sanex shampoo (ground truth HDPE) 9. Plastic cup (ground truth PS) 10. Water bottle cap (ground truth HDPE) 11. Tofu container (ground truth PP) 12. Transparent food bag (suspected LDPE) 13. Shopping bag (ground truth LDPE) 14. Tupperware (ground truth PP) 15. White cork (ground truth PS) 16. Straws (16A - dark straws suspected HDPE / 16B - light straws suspected PP) More information about data capture: Morales, A., Horstrand, P., Guerra, R., Leon, R., Ortega, S., Díaz, M., ... & Sarmiento, R. (2022). Laboratory hyperspectral image acquisition system setup and validation. Sensors, 22(6), 2159.

This repository contains the liquid chromatography–high-resolution mass spectrometry (LC-HRMS) data supporting a non-target screening study of chemical changes associated with polyvinyl chloride (PVC) pipe exposure. Simulated drinking water (DW), conventional mineral hydroponic fertilizer solution (CON), and organic fish-emulsion fertilizer solution (ORG) were analyzed before and after 24-hour exposure to Schedule 40 PVC pipe. The dataset comprises one ultrapure-water blank, one pre-exposure control for each matrix, and three post-exposure biological replicates per matrix, for a total of 13 samples. Data were acquired in positive (PI) and negative (NI) electrospray ionization modes. The repository includes the LC-HRMS data files and two accompanying Excel workbooks. “Raw Data.xlsx” provides retention time, mass-to-charge ratio (m/z), putative metabolite annotations, and individual peak-intensity values for all 13 samples, organized by ionization mode. “Raw Data with SMILES and standard names.xlsx” provides standardized compound names, descriptive sample labels, averaged post-exposure replicate intensities, PubChem compound identifiers (CIDs), SMILES, connectivity SMILES, and InChIKeys for 369 PI and 279 NI annotated features. Compound annotations were assigned through spectral-library matching and should be regarded as putative rather than confirmed identities unless independently validated using authentic standards. Reported peak intensities represent relative instrumental responses and should not be interpreted as absolute compound concentrations.