ibbi_yolov11_oc

Model Card: yolov11x_bb_multi_class_detect_model

Model Details

• Model Name: yolov11x_bb_multi_class_detect_model
• Model Type: Multi-Class Object Detection and Classifier
• Description: This model is designed to detect and classify specific species and genera of bark beetles from images. Unlike single-class models, it has been fine-tuned on a labeled dataset of bark beetle species.

---

Evaluation Datasets

To understand the model's capabilities, its performance was tested on two different types of datasets:

• In-Distribution (ID): This dataset contains images of species the model was trained on. Performance on this dataset shows how well the model identifies familiar species.
• Out-of-Distribution (OOD): This dataset contains images of species that are intentionally different from the training data. Performance here tests the model's ability to handle novel species.

---

Performance

Object Detection & Classification

The model's performance is measured by its mean Average Precision (mAP). This score reflects the model's accuracy in both locating the beetle (bounding box) and assigning the correct species or genus label.

Species-Level Performance

This evaluates the model's ability to identify individual species.

<br> <details> <summary><b>Click to see Per-Species Performance (ID Dataset)</b></summary>

The following list is sorted by Average Precision (AP) from lowest to highest to highlight the most challenging species for the model to identify.

</details>

---

Genus-Level Performance

This evaluates the model's ability to identify the genus, a broader taxonomic rank than species.

<br> <details> <summary><b>Click to see Per-Genus Performance (ID and OOD Datasets)</b></summary>

The following lists are sorted by Average Precision (AP) from lowest to highest to highlight the most challenging genera for the model to identify.

In-Distribution (ID) Genus Performance

<br>

Out-of-Distribution (OOD) Genus Performance

</details>

---

Feature Extraction (Embedding Performance)

The quality of the model's learned feature representations (embeddings) is evaluated by how well they group similar species together.

Internal Cluster Validation

These metrics measure the quality of the clusters formed by the embeddings without referring to ground-truth labels.

External Cluster Validation

These metrics evaluate the clustering performance by comparing the results to the true species labels.

Conclusion: The high external validation scores for the ID dataset show that the model's feature representations are effective at separating the different species it was trained on. This is a significant improvement over single-class or zero-shot models.

Phylogenetic Correlation (Mantel Test)

This test determines if the model's learned features correlate with the evolutionary relationships (phylogeny) between different bark beetle species.