Novel method to automatize flash point detection in small volumes of liquid by computer vision using thermal images
Bibliographic record
Abstract
• Novel methodology for flash point detection of 15 to 300 μL of liquid. • Closed-cup flash point apparatus combined with thermal imaging camera. • Use of computer vision for flash point detection. • Flash point detection at 95 % accuracy. A novel methodology to automate flash point detection of small solvent volumes has been successfully demonstrated and optimized. Flash point temperatures were measured using a closed-cup rapid flash point tester which was paired with a thermal imaging camera. The thermal imaging camara analyses the apparent temperature of the flame before, during and after the opening of the chamber. Two flash point standards and n-eicosane were used to validate the apparatus. Sample volume ranged between 15 and 300 μL which is of interest for the analysis of expensive or harmful liquids. This approach could eventually be extended to measuring flash points in gel polymer electrolytes, for which flammability testing is of significant interest for battery R&D. In this work, 462 flashpoints were collected to feed the machine learning algorithms. Convolutional neural network, support vector machine and random forest algorithms were used to determine the presence/absence of a flame. Flash points were predicted with an accuracy of 95 % and a precision of ±2 °C. Precision was found to be limited by the flash point detector rather than the analysis by computer vision. Other factors such as humidity (22 % to 55 %), atmospheric pressure (between 99.6 to 102.0 kPa) and volume of solvent were found to have little to no influence on flash point detection. The flash point temperatures tested in this study are limited to a range between 50 °C and 176 °C. Regression algorithms were employed to estimate the flash point temperature based on a single measurement presenting an improvement as accurate flash point detection traditionally requires several measurements.
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 | 0.001 | 0.001 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.002 | 0.001 |
| Science and technology studies | 0.000 | 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) | 0.003 | 0.002 |
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".