Skip to content

Latest commit

 

History

History
285 lines (209 loc) · 9.15 KB

README.md

File metadata and controls

285 lines (209 loc) · 9.15 KB
Raw

hive-py

Install

Method: PyPi

pip install hivemapper-python

Method: Clone locally and run

pip install -r requirements.txt

Notes & Limitations

  • Generate a base64 encoded string like 'my-user-name:{apiKey}' to use as input for authorization
  • The Imagery API demo restricts queries to Polygons with a maximum area of 1 km^2
    • This wrapper supports automatically breaking up large geometries into smaller geometries behind the scenes

Usage

Imagery CLI

> python -m imagery.query
usage: query.py [-h] -i INPUT_FILE [-s START_DAY] [-e END_DAY] [-L] [-x] [-d MAX_DIST] [-l MAX_LAG] [-z MAX_ANGLE] -o OUTPUT_DIR [-g] [-w WIDTH] [-M]
                [-I CUSTOM_ID_FIELD] [-S CUSTOM_MIN_DATE_FIELD] [-k] [-E] [-K SKIP_GEO_FILE] [-P IMAGE_POST_PROCESSING] -a AUTHORIZATION [-c NUM_THREADS] [-v]
                [-C] [-b] [-N]

options:
  -h, --help            show this help message and exit
  -i INPUT_FILE, --input_file INPUT_FILE
  -sg SEGMENT_IDS, --segment_ids SEGMENT_IDS
  -s START_DAY, --start_day START_DAY
  -e END_DAY, --end_day END_DAY
  -L, --latest
  -x, --stitch
  -d MAX_DIST, --max_dist MAX_DIST
  -l MAX_LAG, --max_lag MAX_LAG
  -z MAX_ANGLE, --max_angle MAX_ANGLE
  -o OUTPUT_DIR, --output_dir OUTPUT_DIR
  -Z, --zip_dirs
  -Zio, --zip_images_only
  -g, --export_geojson
  -w WIDTH, --width WIDTH
  -m MOUNT, --mount MOUNT
  -M, --merge_metadata
  -I CUSTOM_ID_FIELD, --custom_id_field CUSTOM_ID_FIELD
  -S CUSTOM_MIN_DATE_FIELD, --custom_min_date_field CUSTOM_MIN_DATE_FIELD
  -SF CUSTOM_MIN_DATE_FORMATTING --custom_min_date_formatting CUSTOM_MIN_DATE_FORMATTING
  -Io CUSTOM_OUTPUT_DIR_FIELD, --custom_output_dir_field CUSTOM_OUTPUT_DIR_FIELD
  -Ib CUSTOM_OUTPUT_SUCCESS_FIELD, --custom_output_success_field CUSTOM_OUTPUT_SUCCESS_FIELD
  -Is CUSTOM_OUTPUT_DATE_FIELD, --custom_output_date_field CUSTOM_OUTPUT_DATE_FIELD
  -tI, --track_by_custom_id
  -p, --passthrough_csv_output
  -k, --camera_intrinsics
  -E, --update_exif
  -K SKIP_GEO_FILE, --skip_geo_file SKIP_GEO_FILE
  -P IMAGE_POST_PROCESSING, --image_post_processing IMAGE_POST_PROCESSING
  -a AUTHORIZATION, --authorization AUTHORIZATION
  -c NUM_THREADS, --num_threads NUM_THREADS
  -v, --verbose
  -C, --cache
  -b, --use_batches
  -N, --skip_cached_frames

Bursts CLI

> python -m burts.query
usage: query.py [-h] -i INPUT_FILE -a AUTHORIZAITON

options:
  -h, --help            show this help message and exit
  -i INPUT_FILE, --input_file INPUT_FILE
  -a AUTHORIZATION, --authorization AUTHORIZATION
  -v, --verbose

Python API

Query and download

from imagery import download_files, query_frames

# make the API call to query available data
frames = query_frames(geojson_file, start_day, end_day, output_dir, authorization)

# download the content into folders grouped by its session id
download_files(frames, output_dir)

Create Honey Burst

from bursts import create_bursts

# make the API call to create new bursts by given geojson polygons
# requires polygon type for each of the location
burst_results = create_burts(geojson_file, authorization)

Example

Query imagery for a GeoJSON Polygon Feature

python -m imagery.query -v -M --input_file "test_feature.json" --start_day "2023-07-28" --end_day "2023-07-28" --output_dir "temp" --authorization <your encoded key string>

Query imagery for a OSM road segment

python -m imagery.query -v -M -sg 089283082abbffff0423fcc946ad8fec --start_day "2024-07-28" --end_day "2024-07-28" --output_dir "temp" --authorization <your encoded key string>

Query imagery for multiple OSM road segment

python -m imagery.query -v -M -sg 089283082abbffff0423fcc946ad8fec 088283082abfffff0467f4b6b725f9af --start_day "2024-07-28" --end_day "2024-07-28" --output_dir "temp" --authorization <your encoded key string>

Query imagery for a GeoJSON Polygon Feature, use cache for resumable, use batches

python -m imagery.query -v -M -C -b --input_file "test_feature.json" --start_day "2023-07-28" --end_day "2023-07-28" --output_dir "temp" --authorization <your encoded key string>

Query imagery for a GeoJSON Polygon FeatureCollection; stitch together; save a GeoJSON of LineStrings

python -m imagery.query -v -M -x -g --input_file "test_feature_col.json" --start_day "2023-07-28" --end_day "2023-07-28" --output_dir "temp" --authorization <your encoded key string>

Query imagery for a GeoJSON Polygon FeatureCollection; save a GeoJSON of points

python -m imagery.query -v -M -g --input_file "test_feature_col.json" --start_day "2023-07-28" --end_day "2023-07-28" --output_dir "temp" --authorization <your encoded key string>

Query imagery for a GeoJSON Polygon Feature, add camera intrinsics and encode to exif

python -m imagery.query -v -M -k -E --input_file "test_feature.json" --start_day "2023-07-28" --end_day "2023-07-28" --output_dir "temp" --authorization <your encoded key string>

Focal Length is encoded in pixel units (i.e., not mm) Lens is encoded as <k1> <k2>

Note: exiftool is required to be installed (see https://exiftool.org/)

Converting .shp to Hivemapper-optimized GeoJSON

> python -m util.geo -h
usage: geo.py [-h] [-s SHAPEFILE] [-c CSVFILE] -o OUTPUT_JSON [-w WIDTH] [-I CUSTOM_ID_FIELD] [-S CUSTOM_MIN_DATE_FIELD] [-q]

options:
  -h, --help            show this help message and exit
  -s SHAPEFILE, --shapefile SHAPEFILE
  -c CSVFILE, --csvfile CSVFILE
  -o OUTPUT_JSON, --output_json OUTPUT_JSON
  -w WIDTH, --width WIDTH
  -I CUSTOM_ID_FIELD, --custom_id_field CUSTOM_ID_FIELD
  -S CUSTOM_MIN_DATE_FIELD, --custom_min_date_field CUSTOM_MIN_DATE_FIELD
  -q, --quiet

Skipping last output frames areas

> python -m util.geo -h
python -m imagery.query -v -M -g --input_file "test_feature_col.json" --start_day "2023-07-28" --end_day "2023-07-28" --output_dir "temp" --authorization <your encoded key string> -K last_out/frames.geojson

Skipping multiple output frames areas

> python -m util.geo -h
python -m imagery.query -v -M -g --input_file "test_feature_col.json" --start_day "2023-07-28" --end_day "2023-07-28" --output_dir "temp" --authorization <your encoded key string> -K last_out/frames.geojson,another_out/frames.geojson

Querying API Usage

usage: info.py [-h] -a AUTHORIZATION [-b] [-l LIMIT] [-t] [-v]

options:
  -h, --help            show this help message and exit
  -a AUTHORIZATION, --authorization AUTHORIZATION
  -b, --balance
  -l LIMIT, --limit LIMIT
  -t, --history
  -v, --verbose

Querying Remaining API Credit Balance

python -m account.info -ba <your encoded key string>

Querying API Transaction history (default limit of 25)

python -m account.info -ta <your encoded key string>

Restitching

usage: stitching.py [-h] [-R RESTITCH] [-o OUT] [-d MAX_DIST] [-l MAX_LAG] [-z MAX_ANGLE] [-m MIN_SEQ_SIZE] [-v]

options:
  -h, --help            show this help message and exit
  -R RESTITCH, --restitch RESTITCH
  -o OUT, --out OUT
  -d MAX_DIST, --max_dist MAX_DIST
  -l MAX_LAG, --max_lag MAX_LAG
  -z MAX_ANGLE, --max_angle MAX_ANGLE
  -m MIN_SEQ_SIZE, --min_seq_size MIN_SEQ_SIZE
  -v, --verbose

Restitch a directory out (creates hard links to images)

python -m util.stitching -R out -o out2 -v

Restitch a directory out, but only keep sequences >= 100m (creates hard links to images)

python -m util.stitching -R out -o out2 -v -m 100

Post Processing

  • Install ImageMagick >=7.0.0
  • Use Python >=3.7
python -m imagery.query -v -M -x -g --input_file "test_feature_col.json" --start_day "2023-07-28" --end_day "2023-07-28" --output_dir "temp" --authorization <your encoded key string> -P clahe-smart-clip

Optical Flow (Image Orientation)

Default Usage:

python optical_flow.py input_dir

All Options:

python optical_flow.py input_dir --unzip --max_corners MAX_CORNERS --num_random_checks NUM_RANDOM_CHECKS --threshold_dxdy_ratio THRESHOLD_DXDY_RATIO --turn_threshold TURN_THRESHOLD

clahe-smart-clip (Contrast Limited Adaptive Histogram Equalization with Smart Clipping)

It's highly recommended to use the module directly in order to preserve the original imagery, as well as to tune values for your own purposes.

By default, settings are naively configured to sacrifice aesthetics to improve unsupervised feature detection. Some general deep learning inference use cases and human in the loop use cases may also see benefits from these default settings.

Mitigating Direct Sunlight

directsun clahe1

Mitigating Heavy Shadows

dark1 clahe2

Examples