Non-destructive assessment of chicken egg fertility using hyperspectral imaging technique
Bibliographic record
Abstract
The Canadian chicken industry is a huge one with about 2,836 regulated producers spread across the provinces producing, and of which 61% of production originated from Quebec and Ontario. According to the Agriculture and Agri-food Canada report 2017, total hatching egg set (for both egg production chicks and broilers) was over 1.0 billion. With fertility rate observed in the year 2017 to be around 82%, there were about 180 million unhatched eggs incubated in Canada for year 2017 alone. This meant a whooping sum of at least 311 million Canadian dollars was wasted by the hatchery industries towards incubating unhatched eggs for the year 2017. Whereas, this non-hatching, non-fertile eggs can find useful applications as commercial table eggs or low-grade food stock if they can be detected early and isolated accordingly, especially prior to incubation. The primary goal of this research is to investigate the use of a near infrared (NIR) hyperspectral imaging (HSI) technique in a non-destructive assessment of early chicken egg fertility recognition and discrimination.The first study examined the suitability of a chemometric partial least square (PLS) regression algorithm, towards building a robust model for objective prediction of chicken egg fertility. For the brown eggs on considered incubation days 0 to 4, true positive rates (TPR) ranged from 95.65% to 100% and true negative rates (TNR) ranged from 88.10% to 93.57%. White eggs on the other hand has true positive rates (TPR) ranging from 95.24% to 100% and true negative rates (TNR) ranging from 91.35% to 95.83%. All results were obtained at selected threshold values of between 0.50-0.85. The results indicated that the adapted PLS regression technique can discriminate between fertile and non-fertile eggs, prior to incubation and on different days of incubation. The results were promising with the use of many PLS components (PCs), but the use of fewer PCs shifted classification accuracies in favour of the prevalent class due to the imbalance data structure phenomenon. It therefore became imperative to improve on the present implementation mode of PLS for classification algorithm. Based on the present results, the second study tested the appropriateness of a PLSDA feature selection algorithm, for identifying informative features, towards improving model performance for early chicken egg fertility classification. With only a maximum number of 5 PCs considered, classifier performance greatly improved with selected ratio features; having TPR, TNR, and AUC (area under ROC curve) values in the range of 90-100%. Chicken egg fertility model structure was eventually successfully developed, validated, and verified using optimum number of 3 PCs.Understanding that the modelling approach used to identify informative variables might not be the best approach to translate the identified features into Industrial practice, 10 different classifier performances were compared and contrasted in the third study for adoptability potentials towards building an industrial online chicken egg fertility assessment system. From the sensitivity, specificity, precision, and F1-score values of 100.00%, 87.00%, 93.80%, and 96.80% respectively for brown eggs and 100.00%, 71.40%, 87.80%, and 93.50% respectively for white eggs, the k-nearest neighbours (KNN) classifier was adjudged preferable above its other counterparts. The final study examined the performance of a synthetic minority oversampling technique (SMOTE) algorithm on a larger industrial scale (10, 000) chicken egg fertility data set. KNN classifier already presented as optimal among other classifiers was used for discrimination and performance evaluated from sensitivity (SEN- 92.10%), specificity (SPE- 80.50%), precision (PPV- 99.10%), AUC- 91.70%, and overall accuracy (OVA- 91.60%). Our latest results based on the considered evaluation criteria were comparable with previous results, showing reproducibility potential of our methodology
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 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.001 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
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".