Spatial and Temporally aligned Visible and Infrared UAV images (labelled) and videos (not labelled)
Bibliographic record
Abstract
This dataset was created for sensing of UAVs in the context of the Counter-UAS problem. To implement data fusion methodologies for imaging sensors, namely at pixel level, the data from the visible and infrared sensors must be spatial and temporally aligned. To this end, flight tests were conducted at the University of Victoria's Center for Aerospace Research (UVIC-CfAR) using a TASE 200 gimbal (Visible sensor: SONY FCB-EX1020 PAL, Infrared sensor: FLIR TAU 640 PAL). Additional data collected at the Universidade de Lisboa - Instituto Superior Técnico (IST) using a TeAx ThermalCapture Fusion Zoom was provided. This resulted in two separate sub-datasets: one of labelled UAV images, and one of UAV videos not labelled. All data from Visible and Infrared sensors are spatial and temporally aligned. The labelled dataset includes real frames of a DJI Mavic 2, the VTOL Mini-E (developed at UVIC-CfAR), the hybrid multirotor MIMIQ (developed at UVIC-CfAR), a DJI Mini 3 Pro, and a Zeta FX-61 Phantom Wing and artificial frames of quadcopters, a hexacopter, and a fixed-wing. It includes variety in operational conditions and characteristics, namely range, lighting, blurry and partially cut UAV, presence of birds, and background texture. Images are labelled in the YOLO format. Folders were organized in the YOLO format with 80-10-10 partition for training, test and validation sets. Images were randomly selected for each folder. The dataset of videos that are not labelled includes videos of a DJI Mavic 2, the VTOL Mini-E (developed at UVIC-CfAR), and a DJI Inspire 1. Some videos are in their original unprocessed version. Others are separated into videos of interest, which include the segments with better spatial and temporal alignment and isolation of operational conditions and characteristics.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.001 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.002 | 0.003 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.012 | 0.011 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".