Buffalo Identification in Mixed-Species Environments: A Comparative Deep Learning Approach Using ResNet50 and EfficientNetB3
Bibliographic record
Abstract
Abstract:Background: Buffaloes are integral to agricultural economies, particularly in regions that depend on them for milk production, labor, and income. However, their accurate visual identification in mixed-species environments, especially when co-existing with animals like elephants and rhinos, remains a technological challenge. Methods: This study explores deep learning-based image classification for species-specific buffalo detection using two convolutional neural network architectures: ResNet50 and EfficientNetB3. A balanced image dataset comprising four classes (buffalo, elephant, rhino, zebra) was curated, with training (80%) and validation (20%) splits. The models were fine-tuned using transfer learning, with custom dense layers added atop frozen base layers. EfficientNetB3 used higher-resolution inputs (300x300) and extensive augmentation, while ResNet50 operated on 300x300 images. Performance was evaluated using confusion matrices and key metrics, including validation accuracy, precision, recall, and F1-score, primarily focusing on buffalo classification. Results:ResNet50 achieved a validation accuracy of 47%, and EfficientNetB3 achieved 42%. However, ResNet50 misclassified buffaloes heavily, resulting in a buffalo recall of only 0.07 and an F1-score of 0.11. In contrast, EfficientNetB3 correctly classified 72 out of 200 buffalo images, achieving a buffalo recall of 0.36 and an F1-score of 0.32. These numerical results highlight EfficientNetB3’s superior ability to identify buffaloes accurately in complex visual contexts. Conclusion: EfficientNetB3 is more effective than ResNet50 for buffalo-focused image recognition tasks, offering higher sensitivity and precision in buffalo classification. This study supports the development of AI-powered species-specific monitoring tools, aiding in health tracking, ecological studies, and smart agricultural systems.
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 | 0.002 | 0.002 |
| Meta-epidemiology (narrow) | 0.002 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.002 | 0.001 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.001 | 0.002 |
| Open science | 0.002 | 0.001 |
| Research integrity | 0.002 | 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".