This project automates the preprocessing of RAW agricultural images, particularly those of plants in controlled environments. It takes raw image data, converts it to standard formats (JPG), reconstructs a 3D scene using Structure from Motion (SfM), detects plants (YOLO), refines their locations (bounding boxes), assigns species labels, generates quality control reports, and automatically tracks progress and issues on GitHub. The entire workflow is orchestrated as a pipeline, managed by configuration files for flexibility.
flowchart TD
A0["Pipeline Execution & Orchestration"]
A1["Configuration Management (Hydra)"]
A2["RAW Image Processing & Conversion (RAW to DNG to JPG)"]
A3["Structure from Motion (SfM) Pipeline"]
A4["Plant Detection"]
A5["Bounding Box Processing and Labeling"]
A6["Reporting and Quality Control (QC)"]
A7["Automated Issue Tracking (GitHub)"]
A8["Data Synchronization and Movement"]
A0 -->|Reads configuration| A1
A0 -->|Executes RAW processing| A2
A0 -->|Executes SfM pipeline| A3
A0 -->|Executes plant detection| A4
A0 -->|Executes BBox processing| A5
A0 -->|Executes report generation| A6
A0 -->|Triggers issue creation/update| A7
A0 -->|Executes data synchronization| A8
A1 -->|Provides RAW processing config| A2
A1 -->|Provides SfM parameters| A3
A1 -->|Provides detection settings| A4
A1 -->|Provides BBox parameters| A5
A1 -->|Provides QC/reporting config| A6
A1 -->|Provides sync rules| A8
A2 -->|Provides JPGs for 3D model| A3
A2 -->|Provides JPGs for detection| A4
A3 -->|Provides 3D map for remapping| A5
A4 -->|Provides initial bounding boxes| A5
A5 -->|Provides processed boxes for reports| A6
A8 -->|Stages RAW and model files| A2
- RAW Image Processing & Conversion (RAW -> DNG -> JPG)
- Plant Detection
- Structure from Motion (SfM) Pipeline
- Bounding Box Processing & Labeling
- Pipeline Execution & Orchestration
- Configuration Management (Hydra)
- Data Synchronization & Movement
- Reporting & Quality Control (QC)
- Automated Issue Tracking (GitHub)
Generated by AI Codebase Knowledge Builder