FIELD Database
Field Observation Database - Comprehensive field observations with detailed agricultural context, phenology, and quality control workflows.
Table of contents
- Overview
- Key Features
- Schema Documentation
- 1. Core Identifiers
- 2. Temporal Information
- 3. Geographic & Agricultural Context
- 4. File Paths & Storage
- 5. Image Technical Details
- 6. Detection & Classification
- 7. Complete Taxonomic Information
- 8. Plant Characteristics & Phenology
- 9. Quality Control & Review
- 10. Reference IDs & External Links
- 11. Visualization
- Quality Control Workflow
- External Integrations
- Data Quality Indicators
- Agricultural Context Fields
- Accessing the Data
- Comparison with SEMIF
- Next Steps
Overview
Database Name: FIELD (Field Observation Database)
Purpose: Comprehensive field observation database for plant specimens with detailed agricultural context, taxonomy, quality control workflows, and phenological tracking. Designed for agricultural research and monitoring applications.
Key Features
What Makes FIELD Special: Rich agricultural context, multi-stage quality control, phenological data, dual image formats, and extensive external database linkage.
- Comprehensive Agricultural Context: Detailed crop variety, cover crop, and field condition tracking
- Multi-Stage Quality Control: Initial and final mask review with issue tagging and reviewer tracking
- Rich Phenological Data: Growth stage, reproductive structures, and plant characteristics
- Dual Image Formats: Support for both RAW and JPG image pairs
- Extensive External Linkage: Multiple WIR database rowkey references for cross-system integration
- Temporal Tracking: Multiple timestamps for capture, upload, insertion, and mask modification
Schema Documentation
1. Core Identifiers
Unique identifiers and batch information.
| Field | Type | Description |
|---|---|---|
id | Integer | Primary key, unique record identifier |
alias | String | Alternative or simplified identifier |
batch_id | String | Processing batch identifier |
image_id | String | Unique identifier for the image |
image_index | Integer | Sequential index within a batch or collection |
cutout_name | Float | Name of the associated cutout image |
2. Temporal Information
Complete timestamp tracking for all processing stages.
| Field | Type | Description |
|---|---|---|
camera_datetime | Date | Date and time when image was captured |
db_insert_datetime | Date | Timestamp when record was inserted into database |
upload_datetime_utc | Date | UTC timestamp of image upload |
mask_timestamp | Float | Timestamp of mask creation or modification |
Temporal Analysis: Use these timestamps to analyze processing latency, seasonal patterns, and annotation workflow efficiency.
3. Geographic & Agricultural Context
Field-level environmental and crop information.
| Field | Type | Description |
|---|---|---|
us_state | String | US state where observation was made |
crop_or_fallow | String | Field status: actively cropped or fallow |
crop_type_secondary | String | Secondary crop type classification |
cotton_variety | String | Specific cotton variety if applicable |
cover_crop_family | String | Taxonomic family of cover crop |
cloud_cover | String | Cloud cover conditions during capture |
ground_cover | String | Description of ground cover conditions |
ground_residue | String | Type and amount of residue on ground |
Agricultural Context: This database excels at capturing field-level context that is minimal or absent in other plant databases.
Example agricultural context:
crop_or_fallow: "Cropped"
crop_type_secondary: "Cotton"
cotton_variety: "DP 1646 B2XF"
cover_crop_family: "Fabaceae"
ground_cover: "Moderate"
ground_residue: "High - cotton stalks"
cloud_cover: "Partly cloudy"
4. File Paths & Storage
Locations for images, masks, and various processing stages.
| Field | Type | Description |
|---|---|---|
image_url | String | URL or path to access the image |
developed_image_path | String | Path to processed/developed image |
raw_image_path | String | Path to raw unprocessed image |
cutout_image_path | Float | Path to cutout/cropped image |
final_mask_path | Float | Path to final approved segmentation mask |
final_cutout_path | Float | Path to final cutout image |
extension | String | File extension type |
RAW + JPG Support: Use
has_matching_jpg_and_rawto find image pairs for high-quality analysis.
5. Image Technical Details
Camera metadata and image properties.
| Field | Type | Description |
|---|---|---|
exif_meta | String | EXIF metadata from camera |
height | String | Image height dimension |
size_mib | Float | File size in mebibytes (MiB) |
has_matching_jpg_and_raw | Integer | Boolean flag for paired JPG and RAW files |
is_preprocessed | Integer | Boolean flag indicating preprocessing status |
6. Detection & Classification
Model predictions and confidence scores.
| Field | Type | Description |
|---|---|---|
class_id | Integer | Numeric class identifier |
bbox_xywh | Float | Bounding box coordinates [x, y, width, height] |
det_pred_conf | Float | Detection prediction confidence score (0-1) |
app_species | String | Application-level species identifier |
has_mat_pred | Float | Boolean flag for material/mat prediction |
has_mat_pred_conf | Float | Confidence score for material prediction |
Confidence filtering:
# High confidence detections
det_pred_conf > 0.85
# Very high confidence
det_pred_conf > 0.95
7. Complete Taxonomic Information
Full taxonomic hierarchy with external database codes.
| Field | Type | Description |
|---|---|---|
usda_symbol | String | USDA PLANTS database symbol |
eppo | String | EPPO code |
common_name | String | Common name of the plant |
taxonomic_group | String | High-level taxonomic group |
taxonomic_class | String | Taxonomic class |
taxonomic_subclass | String | Taxonomic subclass |
taxonomic_order | String | Taxonomic order |
taxonomic_family | String | Taxonomic family |
taxonomic_genus | String | Taxonomic genus |
taxonomic_species_name | String | Full species name |
taxonomic_authority | String | Authority citation for taxonomy |
multi_species_USDA_symbol | Float | Multiple species USDA symbols if applicable |
Example taxonomy:
common_name: "Palmer amaranth"
usda_symbol: "AMPA"
eppo: "AMAPA"
taxonomic_family: "Amaranthaceae"
taxonomic_genus: "Amaranthus"
taxonomic_species_name: "Amaranthus palmeri"
taxonomic_authority: "S. Watson"
8. Plant Characteristics & Phenology
Growth stages, reproductive structures, and morphology.
| Field | Type | Description |
|---|---|---|
plant_type | String | General plant type classification |
growth_habit | String | Growth habit (forb, grass, shrub, vine, tree) |
duration | String | Life cycle duration (annual, biennial, perennial) |
growth_stage | String | Current phenological growth stage |
flower_fruit_or_seeds | Integer | Boolean flag for presence of reproductive structures |
stem | String | Stem characteristics description |
size_class | String | Categorical size classification |
Phenological Research: These fields enable detailed growth stage tracking and reproductive phenology studies.
Growth stages may include:
- Seedling / Cotyledon
- Vegetative
- Bolting
- Flowering
- Seed set
- Senescence
Example phenological data:
growth_stage: "Flowering"
flower_fruit_or_seeds: 1
duration: "Annual"
growth_habit: "Forb"
size_class: "Large"
9. Quality Control & Review
Multi-stage annotation workflow tracking.
| Field | Type | Description |
|---|---|---|
mask_status | Float | Status of mask review/approval |
mask_reviewer | Float | Identifier of person who reviewed mask |
initial_mask_issue_tag | Float | Tag for issues found in initial mask |
final_mask_issue_tag | Float | Tag for issues in final mask |
refine_params | Float | Parameters used for mask refinement |
processing_note | Float | Notes about processing steps or issues |
tags | Float | General tags for categorization or flagging |
category_note | Float | Notes specific to category assignment |
Quality Workflow: The FIELD database implements a comprehensive 5-stage quality control pipeline (see below).
10. Reference IDs & External Links
Integration with WIR (Weed Image Repository) system.
| Field | Type | Description |
|---|---|---|
wirmaster_ref_id | String | Reference ID to master WIR database |
wirmastermeta_rowkey | String | Row key for WIR master metadata table |
wircovercropsmeta_rowkey | String | Row key for WIR cover crops metadata |
wircropsmeta_rowkey | String | Row key for WIR crops metadata |
wirimagerefs_rowkey | String | Row key for WIR image references |
wirweedsmeta_rowkey | String | Row key for WIR weeds metadata |
link | String | General reference link or URL |
WIR Integration provides cross-references to:
- Master metadata records
- Crop-specific information
- Cover crop details
- Weed species data
- Image references
11. Visualization
Display properties for rendering.
| Field | Type | Description |
|---|---|---|
rgb | String | RGB color values for visualization |
Quality Control Workflow
The FIELD database supports a comprehensive quality control pipeline:
Stage 1: Initial Processing
- Field:
is_preprocessed - Notes:
processing_note - Action: Basic image processing and validation
Stage 2: Initial Mask Review
- Field:
initial_mask_issue_tag - Reviewer:
mask_reviewer - Action: First pass quality check and issue identification
Stage 4: Final Review
- Field:
final_mask_issue_tag - Status:
mask_status - Action: Final quality approval
Stage 5: Approval & Export
- Fields:
final_mask_path,final_cutout_path - Action: Generation of approved final outputs
External Integrations
USDA PLANTS Database
Access via usda_symbol
EPPO Database
Access via eppo code
WIR System
Extensive cross-database linking
Multiple rowkey references
Data Quality Indicators
Use these fields to filter for high-quality data:
| Indicator | Field | Recommended Value |
|---|---|---|
| Detection Confidence | det_pred_conf | > 0.85 (high confidence) |
| Material Confidence | has_mat_pred_conf | > 0.8 (if applicable) |
| Mask Status | mask_status | Approved/completed |
| Image Format | has_matching_jpg_and_raw | 1 (paired formats) |
| File Size | size_mib | Check for reasonable values |
| Reviewer | mask_reviewer | Not null (human reviewed) |
Agricultural Context Fields
Unique Strength: The FIELD database excels at capturing agricultural context that is minimal or absent in other plant databases.
What’s Captured:
✅ Crop Information
- Primary and secondary crop types
- Specific varieties (e.g., cotton varieties)
- Cover crop families
✅ Field Status
- Active cropping vs. fallow
- Crop rotation information
✅ Environmental Conditions
- Cloud cover during capture
- Ground cover descriptions
- Residue type and amount
✅ Geographic Context
- US state location
- Field-level coordinates
Accessing the Data
Query this database using the AgIR-CVToolkit. The toolkit supports filtering by crop type, phenology, quality metrics, and more.
Learn How to Query FIELD → View Complete Query Guide →
Comparison with SEMIF
| Feature | SEMIF | FIELD |
|---|---|---|
| Focus | ML training | Agricultural research |
| Records | Individual cutouts | Field specimens |
| Attributes | 62 | 72 |
| Primary Use | Object detection | Context analysis |
| Annotations | Bounding boxes | Phenology + context |
| Quality Control | Automated metrics | Multi-stage human review |
| Agricultural Context | Minimal | Extensive |
| External Links | USDA, EPPO | USDA, EPPO, WIR system |
When to Use: Choose SEMIF for training object detection models. Choose FIELD for agricultural research requiring crop context and phenological data.
Next Steps
📊 View Statistics
Explore phenological and geographic distributions
🖼️ See Examples
Browse field observations with context
🔧 Access the Data
Learn how to query with AgIR-CVToolkit
🤖 Compare with SEMIF
See the ML-optimized database