ibbi 0.1.3

A package for bark and ambrosia beetle identification.

Intelligent Bark Beetle Identifier (IBBI)

IBBI is a Python package that provides a simple and unified interface for detecting and classifying bark and ambrosia beetles from images using state-of-the-art computer vision models.

This package is designed to support entomological research by automating the laborious task of beetle identification, enabling high-throughput data analysis for ecological studies, pest management, and biodiversity monitoring.

Motivation

The ability to accurately identify bark and ambrosia beetles is critical for forest health and pest management. However, traditional methods face significant challenges:

• They are slow and time-consuming.
• They require highly specialized expertise.
• They create a bottleneck for large-scale research.

The IBBI package provides a powerful, modern solution to overcome these obstacles by using pre-trained, open-source models to automate detection and classification from images, lowering the barrier to entry for researchers.

Key Features

• Simple API: Access powerful detection and classification models with a single function call: ibbi.create_model().
• Multiple Model Types:
  • Single-Class Detection: Detect the presence of any beetle in an image.
  • Multi-Class Species Detection: Identify the species of a beetle from an image.
  • Zero-Shot Detection: Detect objects using a text prompt (e.g., "insect"), without prior training on that specific class.
• Pre-trained Models: Leverages pre-trained models hosted on the Hugging Face Hub for immediate use.
• Model Explainability: Understand model predictions using SHAP (SHapley Additive exPlanations) to visualize which parts of an image contribute to the identification.
• Extensible: Designed to easily incorporate new model architectures in the future.

---

Table of Contents

• Intelligent Bark Beetle Identifier (IBBI)
  • Motivation
  • Key Features
  • Table of Contents
  • Workflow: How the Models Were Built
  • Package API and Usage
    • Usage Examples
  • Installation
  • Quick Start
  • Advanced Usage
    • Feature Extraction
    • Model Explainability with SHAP
  • Available Models
  • How to Contribute
  • License

---

Workflow: How the Models Were Built

The models in ibbi are the result of a comprehensive data collection, annotation, and training pipeline.

1. Data Collection and Curation: The process begins with data collection from various sources. A zero-shot detection model performs initial beetle localization, followed by human-in-the-loop verification to ensure accurate bounding box annotations.
2. Model-Specific Training Data: The annotated dataset is curated for different model types:
   • Single-Class Detection: Models are trained on all images with verified beetle localizations.
   • Multi-Class Species Detection: Models are trained only on images with both verified localizations and species-level labels. To ensure robustness, species with fewer than 50 images are excluded.
3. Evaluation and Deployment: A held-out test set is used to evaluate all models. Performance metrics can be viewed with ibbi.list_models(). Final models are deployed to the Hugging Face Hub for easy access.

---

Package API and Usage

The ibbi package is designed to be simple and intuitive.

The main components are:

• Inputs: Images (file paths, URLs, PIL/numpy objects) and a model name string.
• model.predict(): The main prediction function. Returns bounding boxes, scores, and class labels.
• model.extract_features(): Extracts deep feature embeddings from images for downstream tasks.
• Helper Functions: get_dataset() to load the training/test data, list_models() to see available models, and explain_with_shap() to visualize model predictions using SHAP.

Usage Examples

Here are visual examples of what you can do with ibbi.

Input Image	Single-Class Detection (yolov10x_bb_detect_model)	Multi-Class Species Detection (yolov10x_bb_multi_class_detect_model)	Zero-Shot Detection (grounding_dino_detect_model)

---

Installation

This package requires PyTorch. For compatibility with your specific hardware (e.g., CUDA-enabled GPU), please install PyTorch before installing ibbi.

1. Install PyTorch

Follow the official instructions at pytorch.org to install the correct version for your system.

2. Install IBBI

Once PyTorch is installed, install the package from PyPI:

pip install ibbi

---

Quick Start

Using ibbi is straightforward. Load a pre-trained model and immediately use it for inference.

import ibbi
from PIL import Image

# Use a URL or local path
image_source = "IBBI/docs/assets/images/beetles.png"

# --- 1. Load All Model Types ---
species_detector = ibbi.create_model("yolov10x_bb_multi_class_detect_model", pretrained=True)
detector = ibbi.create_model("yolov10x_bb_detect_model", pretrained=True)
zs_detector = ibbi.create_model("grounding_dino_detect_model", pretrained=True)

# --- 2. Run Prediction for Each Model ---
species_results = species_detector.predict(image_source)
detection_results = detector.predict(image_source)
zs_results = zs_detector.predict(image_source, text_prompt="insect")

# --- 3. Print Results with Name and Confidence ---

# Example for Multi-Class Species Detection
print("--- Multi-Class Species Detection ---")
for box in species_results[0].boxes:
    class_name = species_results[0].names[int(box.cls)]
    confidence = float(box.conf)
    print(f"  - Detected '{class_name}' with confidence {confidence:.2f}")

# Example for Single-Class Detection
print("\n--- Single-Class Detection ---")
for box in detection_results[0].boxes:
    class_name = detection_results[0].names[int(box.cls)]
    confidence = float(box.conf)
    print(f"  - Detected '{class_name}' with confidence {confidence:.2f}")

# Example for Zero-Shot Detection
print("\n--- Zero-Shot Detection (prompt: 'insect') ---")
for box in zs_results[0].boxes:
    class_name = zs_results[0].names[int(box.cls)]
    confidence = float(box.conf)
    print(f"  - Detected '{class_name}' with confidence {confidence:.2f}")

# To visualize the bounding boxes, you can uncomment the following lines:
# species_results[0].show()
# detection_results[0].show()
# zs_results[0].show()

For more detailed, hands-on demonstrations, please see the example notebooks located in the notebooks/ folder of the repository.

---

Advanced Usage

Feature Extraction

All models can extract deep feature embeddings from an image. These vectors are useful for downstream tasks like clustering or similarity analysis.

# Assuming 'species_detector' is a loaded model
features = species_detector.extract_features(image_source)
print(f"Extracted feature vector shape: {features.shape}")

Model Explainability with SHAP

Understand why a model made a certain prediction using SHAP. This is crucial for building trust and interpreting the model's decisions by highlighting which pixels were most influential.

import ibbi

# Load a model
model = ibbi.create_model("yolov10x_bb_multi_class_detect_model", pretrained=True)

# Get images to explain and a background dataset
# This can be any dataset the model is applied to
explain_data = ibbi.get_dataset(split="train", streaming=True).take(5)
background_data = ibbi.get_dataset(split="train", streaming=True).skip(5).take(10)

# Generate explanations (this is computationally intensive)
shap_explanation = ibbi.explain_with_shap(
    model=model,
    explain_dataset=list(explain_data),
    background_dataset=list(background_data),
    num_explain_samples=1,
    num_background_samples=5
)

# Plot the explanation for the first image
ibbi.plot_shap_explanation(shap_explanation[0], model)

---

Available Models

To see a list of available models and their performance metrics directly from Python, run:

import ibbi

# Returns a pandas DataFrame
models_df = ibbi.list_models(as_df=True)
print(models_df)

A detailed list can also be found in the src/ibbi/data/ibbi_model_summary.csv file.

---

How to Contribute

Contributions are welcome! If you would like to improve IBBI, please see the Contributing Guide.

License

This project is licensed under the MIT License. See the LICENSE file for details.