SemiF-PlantDetection

This project is designed to update the detection models used for detecting plants in SemiField (AgIR) data. It works by preparing data selected from a database and long-term storage, which can optionally include sending it for human annotation via CVAT. The cleaned and structured data is then used to train an object detection model (like YOLO) to recognize plants, all managed by a flexible configuration system (hydra).

flowchart TD
    A0["Hydra Configuration System
"]
    A1["Pipeline Modes
"]
    A2["Data Selection from Database
"]
    A3["Image Retrieval
"]
    A4["CVAT Data Preparation & Import
"]
    A5["Training Data Structuring
"]
    A6["Model Training
"]
    A7["Core Utility Functions
"]
    A8["Data and Secrets Locations
"]
    A0 -- "Selects Mode" --> A1
    A0 -- "Configures Paths" --> A8
    A0 -- "Provides Config" --> A2
    A0 -- "Provides Config" --> A3
    A0 -- "Provides Config" --> A4
    A0 -- "Provides Config" --> A5
    A0 -- "Provides Config" --> A6
    A1 -- "Executes Task" --> A2
    A1 -- "Executes Task" --> A3
    A1 -- "Executes Task" --> A4
    A1 -- "Executes Task" --> A5
    A1 -- "Executes Task" --> A6
    A2 -- "Reads/Writes Data" --> A8
    A3 -- "Reads/Writes Images" --> A8
    A3 -- "Uses Utilities" --> A7
    A4 -- "Writes Data" --> A8
    A4 -- "Uses Utilities" --> A7
    A5 -- "Reads/Writes Data" --> A8
    A5 -- "Uses Utilities" --> A7
    A6 -- "Writes Model" --> A8
    A6 -- "Uses Utilities" --> A7
    A7 -- "Accesses Resources" --> A8

Loading

Chapters

1. Hydra Configuration System
2. Pipeline Modes
3. Data and Secrets Locations
4. Data Selection from Database
5. Image Retrieval
6. CVAT Data Preparation & Import
7. Training Data Structuring
8. Model Training
9. Core Utility Functions

---

Generated by AI Codebase Knowledge Builder