Real-time and resource-efficient banana bunch detection and localization with YOLO-BRFB on edge devices
Bibliographic record
Abstract
Reliable detection and spatial localization of banana bunches are essential prerequisites for the development of autonomous harvesting technologies. Current methods face challenges in achieving high detection accuracy and efficient deployment due to their structural complexity and significant computational demands. This study proposes YOLO-BRFB, a lightweight and precise system designed for detection and 3D localization of bananas in orchard environments. First, the YOLOv8 framework is improved by integrating the BasicRFB module, enhancing feature extraction for small targets and cluttered backgrounds while reducing model complexity. Then, a binocular vision system is used for localization, estimating 3D spatial coordinates with high accuracy and ensuring robust performance under diverse lighting and occlusion conditions. Finally, the system is optimized for edge-device deployment, achieving real-time processing with minimal computational resources. Experimental results demonstrate that YOLO-BRFB achieves a precision of 0.957, recall of 0.922, mAP of 0.961, and F1-score of 0.939, surpassing YOLOv8 in both recall and mAP. The average positioning error of the system along the X-axis is 12.33 mm, the average positioning error along the Y-axis is 11.11 mm, and the average positioning error along the Z-axis is 16.33 mm. The system has an inference time of 8.6 milliseconds on an Nvidia Orin NX with a GPU memory requirement of 1.7 GB. This study is among the first to focus on a lightweight approach optimized for deployment on edge computing devices. These results highlight the practical applicability of YOLO-BRFB in real-world agricultural scenarios, providing a cost-effective solution for precision harvesting.
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 | 0.000 |
| Bibliometrics | 0.001 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.000 |
| 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".