WilData Scripts Reference

This page documents all batch scripts available in the WilData package for data management operations.

Overview

All scripts are located in wildata/scripts/ directory.

Quick Reference

Script	Purpose	Config File
import-dataset-example.bat	Import single dataset	configs/import-config-example.yaml
bulk-import-dataset.bat	Bulk import datasets	configs/bulk-import-*.yaml
create-roi-dataset.bat	Create ROI dataset	configs/roi-create-config.yaml
bulk-roi-create-config.bat	Bulk create ROI datasets	configs/bulk-roi-create-config.yaml
update-gps-example.bat	Update GPS from CSV	configs/gps-update-config-example.yaml
visualize_data.bat	Visualize dataset	None
dvc-setup.bat	Setup DVC	None
launch_api.bat	Launch REST API	.env
running_tests.bat	Run tests	None

---

Data Import Scripts

import-dataset-example.bat

Purpose: Import a single dataset from COCO, YOLO, or Label Studio format.

Location: wildata/scripts/import-dataset-example.bat

Command:

uv run wildata import-dataset --config configs\import-config-example.yaml

Configuration: wildata/configs/import-config-example.yaml

Key Parameters:

source_path: "path/to/annotations.json"
source_format: "coco"  # coco, yolo, ls
dataset_name: "my_dataset"

root: "data"
split_name: "train"  # train, val, test

transformations:
  enable_bbox_clipping: true
  enable_tiling: true
  tiling:
    tile_size: 800
    stride: 640

roi_config:
  roi_box_size: 384
  random_roi_count: 2

Example Usage:

cd wildata

# Edit config file first
notepad configs\import-config-example.yaml

# Run import
scripts\import-dataset-example.bat

Output: - Master format dataset in data/datasets/ - Processed images (tiled if enabled) - ROI dataset (if configured)

---

bulk-import-dataset.bat

Purpose: Import multiple datasets in batch mode.

Location: wildata/scripts/bulk-import-dataset.bat

Command:

uv run wildata bulk-import-datasets --config configs\bulk-import-config-example.yaml -n 2

Configuration: wildata/configs/bulk-import-train.yaml or bulk-import-val.yaml

Parameters: - -n 2: Number of parallel workers (uses threading on Windows) - --config: Path to bulk import config

Example Config:

# configs/bulk-import-train.yaml
source_paths:
  - "D:/annotations/dataset1.json"
  - "D:/annotations/dataset2.json"
  - "D:/annotations/dataset3.json"

source_format: "coco"
root: "D:/data"
split_name: "train"

# Shared transformation settings
transformations:
  enable_tiling: true
  tiling:
    tile_size: 800
    stride: 640

Example Usage:

cd wildata
scripts\bulk-import-dataset.bat

Features: - Parallel processing (thread-based) - Progress tracking - Error handling per dataset - Summary report

---

ROI Dataset Scripts

create-roi-dataset.bat

Purpose: Create Region of Interest (ROI) classification dataset from detection annotations.

Location: wildata/scripts/create-roi-dataset.bat

Command:

uv run wildata create-roi-dataset --config configs\roi-create-config.yaml

Configuration: wildata/configs/roi-create-config.yaml

Key Parameters:

source_path: "annotations.json"
source_format: "coco"
dataset_name: "roi_dataset"

roi_config:
  roi_box_size: 128        # Size of extracted ROI
  min_roi_size: 32         # Min object size to extract
  random_roi_count: 10     # Background samples per image
  background_class: "background"
  save_format: "jpg"
  quality: 95

Use Cases: - Hard sample mining - Error analysis - Training classification models - Creating balanced datasets

Example Usage:

cd wildata
scripts\create-roi-dataset.bat

Output: - ROI image crops - Classification labels - Class mapping JSON - Statistics file

---

bulk-roi-create.bat

Purpose: Create multiple ROI datasets in batch.

Location: Script not shown, but referenced in configs

Configuration: wildata/configs/bulk-roi-create-config.yaml

Example Config:

source_paths:
  - "dataset1.json"
  - "dataset2.json"

source_format: "coco"
split_name: "val"

roi_config:
  roi_box_size: 128
  random_roi_count: 5

---

GPS Management Scripts

update-gps-example.bat

Purpose: Update image EXIF GPS data from CSV file.

Location: wildata/scripts/update-gps-example.bat

Command:

uv run wildata update-gps-from-csv --config configs\gps-update-config-example.yaml

Configuration: wildata/configs/gps-update-config-example.yaml

Key Parameters:

image_folder: "path/to/images"
csv_path: "gps_coordinates.csv"
output_dir: "output/images"

skip_rows: 0
filename_col: "filename"
lat_col: "latitude"
lon_col: "longitude"
alt_col: "altitude"

CSV Format:

filename,latitude,longitude,altitude
image1.jpg,40.7128,-74.0060,10.5
image2.jpg,40.7589,-73.9851,15.2

Example Usage:

cd wildata

# Prepare CSV with GPS data
# Edit config
notepad configs\gps-update-config-example.yaml

# Run update
scripts\update-gps-example.bat

Output: - Images with updated EXIF GPS - Summary report - Error log (if any)

---

Visualization Scripts

visualize_data.bat

Purpose: Launch FiftyOne visualization for datasets.

Location: wildata/scripts/visualize_data.bat

Command:

uv run wildata visualize-dataset --dataset my_dataset --split train

Example Usage:

cd wildata

# Visualize training set
uv run wildata visualize-dataset --dataset my_dataset --split train

# Or use script
scripts\visualize_data.bat

Features: - Interactive dataset viewer - Annotation visualization - Filtering and search - Statistics display

---

DVC Scripts

dvc-setup.bat

Purpose: Initialize and configure DVC for data versioning.

Location: wildata/scripts/dvc-setup.bat

Command:

# Initialize DVC
dvc init

# Add remote storage
dvc remote add -d myremote <storage_path>

Storage Options:

Local StorageAWS S3Google Cloud

dvc remote add -d local D:\dvc-storage

dvc remote add -d s3remote s3://bucket/path
dvc remote modify s3remote access_key_id YOUR_KEY
dvc remote modify s3remote secret_access_key YOUR_SECRET

dvc remote add -d gcs gs://bucket/path
set GOOGLE_APPLICATION_CREDENTIALS=path\to\credentials.json

Example Usage:

cd wildata
scripts\dvc-setup.bat

# Track data
dvc add data\datasets\my_dataset

# Commit DVC file
git add data\datasets\my_dataset.dvc
git commit -m "Add dataset"

# Push to remote
dvc push

DVC Workflow:

# On another machine
git pull
dvc pull  # Downloads data

---

API Scripts

launch_api.bat

Purpose: Launch WilData REST API server.

Location: wildata/scripts/launch_api.bat

Command:

uv run python -m wildata.api.main

Default Port: 8441

Example Usage:

cd wildata
scripts\launch_api.bat

Access: - API: http://localhost:8441 - Docs: http://localhost:8441/docs - Redoc: http://localhost:8441/redoc

API Endpoints:

Import Dataset

POST /api/v1/datasets/import
Content-Type: application/json

{
  "source_path": "/path/to/data.json",
  "source_format": "coco",
  "dataset_name": "my_dataset",
  "root": "data"
}

List Datasets

GET /api/v1/datasets?root=data

Create ROI Dataset

POST /api/v1/roi/create
Content-Type: application/json

{
  "source_path": "/path/to/data.json",
  "source_format": "coco",
  "dataset_name": "roi_dataset",
  "roi_config": {
    "roi_box_size": 128,
    "random_roi_count": 10
  }
}

Job Status

GET /api/v1/jobs/{job_id}

Environment Variables:

# In .env
WILDATA_API_HOST=0.0.0.0
WILDATA_API_PORT=8441
WILDATA_API_DEBUG=false

Full API Documentation: See WilData API Reference

---

Testing Scripts

running_tests.bat

Purpose: Run WilData test suite.

Location: wildata/scripts/running_tests.bat

Command:

uv run pytest tests/ -v

Example Usage:

cd wildata
scripts\running_tests.bat

Test Categories: - Format adapter tests - Transformation tests - Validation tests - API tests - Integration tests

Run Specific Tests:

# Test imports
uv run pytest tests/test_coco_import.py -v

# Test transformations
uv run pytest tests/test_transformations.py -v

# Test API
uv run pytest tests/api/ -v

# With coverage
uv run pytest --cov=wildata tests/

---

Common Workflows

Dataset Preparation Workflow

# 1. Import dataset
cd wildata
scripts\import-dataset-example.bat

# 2. Visualize
scripts\visualize_data.bat

# 3. Export for training
uv run wildata dataset export my_dataset --format yolo

ROI Extraction Workflow

# 1. Import detection dataset
scripts\import-dataset-example.bat

# 2. Create ROI dataset
scripts\create-roi-dataset.bat

# 3. Visualize ROI dataset
uv run wildata visualize-dataset --dataset roi_dataset

GPS Management Workflow

# 1. Extract GPS from images
# (using WildDetect extract_gps.bat)

# 2. Update GPS if needed
cd wildata
scripts\update-gps-example.bat

# 3. Verify GPS data
# Check EXIF data in images

DVC Workflow

# Setup (once)
cd wildata
scripts\dvc-setup.bat

# After each dataset import
dvc add data\datasets\new_dataset
git add data\datasets\new_dataset.dvc
git commit -m "Add new dataset"
dvc push

# On other machines
git pull
dvc pull

---

Configuration Examples

Complete Import Config

# configs/import-config-example.yaml
source_path: "D:/annotations/dataset.json"
source_format: "coco"
dataset_name: "wildlife_train"

root: "D:/data"
split_name: "train"
processing_mode: "batch"

# Label Studio integration
ls_xml_config: "configs/label_studio_config.xml"
ls_parse_config: false

# ROI extraction
disable_roi: false
roi_config:
  random_roi_count: 2
  roi_box_size: 384
  min_roi_size: 32
  background_class: "background"
  sample_background: true

# Transformations
transformations:
  enable_bbox_clipping: true
  bbox_clipping:
    tolerance: 5
    skip_invalid: false

  enable_tiling: true
  tiling:
    tile_size: 800
    stride: 640
    min_visibility: 0.7
    max_negative_tiles_in_negative_image: 2
    dark_threshold: 0.7

  enable_augmentation: false
  augmentation:
    rotation_range: [-45, 45]
    probability: 1.0
    num_transforms: 2

---

Troubleshooting

Import Fails

Issue: Dataset import fails with validation errors

Solutions: 1. Check source file format is correct 2. Verify all image paths are valid 3. Check bbox coordinates are within image bounds 4. Use --verbose flag for detailed errors

DVC Push Fails

Issue: Can't push data to remote

Solutions: 1. Verify remote credentials 2. Check network connection 3. Verify remote storage path exists 4. Use dvc remote list to check configuration

API Won't Start

Issue: API server fails to start

Solutions: 1. Check port 8441 is not in use 2. Verify .env file configuration 3. Check all dependencies installed 4. Look at error logs

Out of Memory

Issue: Import fails with memory error

Solutions: 1. Use processing_mode: "streaming" 2. Reduce number of parallel workers 3. Process datasets one at a time 4. Disable transformations temporarily

---

Next Steps

• WilData Configuration Reference
• WilData CLI Reference
• Dataset Preparation Tutorial
• WilData API Documentation