fAIr Predictor
Run fAIr model predictions anywhere. Works on CPU, compatible with serverless functions.
Requires Python >= 3.10.
Installation
# core (RAMP / TFLite models)
uv add fairpredictor
# or
pip install fairpredictor
# with YOLO / ONNX support
uv add "fairpredictor[pytorch]"
pip install "fairpredictor[pytorch]"
# with TensorFlow support
uv add "fairpredictor[tensorflow]"
pip install "fairpredictor[tensorflow]"
# everything
uv add "fairpredictor[all]"
pip install "fairpredictor[all]"
Python usage
import asyncio
from predictor import DEFAULT_OAM_TMS_MOSAIC, DEFAULT_RAMP_MODEL, DEFAULT_YOLO_MODEL, predict
bbox = [100.56228021333352, 13.685230854641182, 100.56383321235313, 13.685961853747969]
zoom_level = 20
# RAMP (TFLite) prediction
result = asyncio.run(
predict(
model_path=DEFAULT_RAMP_MODEL,
zoom_level=zoom_level,
tms_url=DEFAULT_OAM_TMS_MOSAIC,
bbox=bbox,
confidence=0.5,
tolerance=0.5,
area_threshold=3,
orthogonalize=True,
)
)
print(result)
# YOLO (ONNX) prediction
result = asyncio.run(
predict(
model_path=DEFAULT_YOLO_MODEL,
zoom_level=zoom_level,
tms_url=DEFAULT_OAM_TMS_MOSAIC,
bbox=bbox,
)
)
print(result)
predict() parameters
| Parameter | Type | Default | Description |
|---|---|---|---|
model_path |
str |
required | URL or local path to model (.tflite, .onnx, .pt, .h5) |
zoom_level |
int |
required | Tile zoom level (18-22) |
tms_url |
str |
OAM mosaic | TMS URL template {z}/{x}/{y} |
bbox |
list[float] |
None |
[min_lon, min_lat, max_lon, max_lat] |
geojson |
dict|str |
None |
GeoJSON polygon as alternative to bbox |
confidence |
float |
0.5 |
Confidence threshold 0-1 |
tolerance |
float |
0.5 |
Polygon simplification tolerance |
area_threshold |
float |
3 |
Minimum polygon area |
orthogonalize |
bool |
True |
Apply orthogonalization |
ortho_skew_tolerance_deg |
int |
15 |
Orthogonalization skew tolerance (0-45) |
ortho_max_angle_change_deg |
int |
15 |
Maximum angle change for orthogonalization (0-45) |
get_predictions_as_points |
bool |
True |
Include centroid points in output |
make_geoms_valid |
bool |
True |
Validate and fix output polygons |
task |
str |
"segmentation" |
Task type (segmentation only for now) |
output_path |
str |
None |
Directory to save outputs; auto-generated if not set |
debug |
bool |
False |
Save intermediate rasters for debugging |
Either bbox or geojson must be provided.
Development
# Install all dependency groups
just install
# Run linting + type checking + tests
just check
# Lint only
just lint
# Tests only
just test
Load testing
Always obtain permission from the server admin before load testing.
uv run locust -f locust.py
Set HOST to the base URL of the predictor API.
Docker
Build
docker build -t fairpredictor .
Run
docker run --rm -p 8000:8000 fairpredictor
API
- Swagger UI: http://localhost:8000/docs
- ReDoc: http://localhost:8000/redoc
Example request
{
"bbox": [100.56228021333352, 13.685230854641182, 100.56383321235313, 13.685961853747969],
"checkpoint": "https://api-prod.fair.hotosm.org/api/v1/workspace/download/ramp/baseline.tflite",
"zoom_level": 20,
"source": "https://apps.kontur.io/raster-tiler/oam/mosaic/{z}/{x}/{y}.png",
"confidence": 50,
"tolerance": 0.5,
"area_threshold": 2,
"orthogonalize": true,
"task": "segmentation"
}
1 maintainer
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
fairpredictor-0.5.1.tar.gz
(10.9 kB
view details)
Built Distribution
Filter files by name, interpreter, ABI, and platform.
If you're not sure about the file name format, learn more about wheel file names.
Copy a direct link to the current filters
File details
Details for the file fairpredictor-0.5.1.tar.gz.
File metadata
- Download URL: fairpredictor-0.5.1.tar.gz
- Upload date:
- Size: 10.9 kB
- Tags: Source
- Uploaded using Trusted Publishing? No
- Uploaded via:
uv/0.11.2 {"installer":{"name":"uv","version":"0.11.2","subcommand":["publish"]},"python":null,"implementation":{"name":null,"version":null},"distro":{"name":"Ubuntu","version":"24.04","id":"noble","libc":null},"system":{"name":null,"release":null},"cpu":null,"openssl_version":null,"setuptools_version":null,"rustc_version":null,"ci":true}
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
cf0f40a04a7ce52fbd67efa0524d522e8688d7d31cd3130c4b092e42af1427fe
|
|
| MD5 |
bb6b5983fbbbf4ba629f11fa658b5fd5
|
|
| BLAKE2b-256 |
3986123eeb34bd823be2a2f642570ce59e717be74ced6bec91d87219b8278385
|
File details
Details for the file fairpredictor-0.5.1-py3-none-any.whl.
File metadata
- Download URL: fairpredictor-0.5.1-py3-none-any.whl
- Upload date:
- Size: 14.0 kB
- Tags: Python 3
- Uploaded using Trusted Publishing? No
- Uploaded via:
uv/0.11.2 {"installer":{"name":"uv","version":"0.11.2","subcommand":["publish"]},"python":null,"implementation":{"name":null,"version":null},"distro":{"name":"Ubuntu","version":"24.04","id":"noble","libc":null},"system":{"name":null,"release":null},"cpu":null,"openssl_version":null,"setuptools_version":null,"rustc_version":null,"ci":true}
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
bbf182df0b80761c8868154ab836c7865a106aff141a835f0f9c6649f87c1556
|
|
| MD5 |
e5db910dc66ff963fc934bc61dde28b8
|
|
| BLAKE2b-256 |
a6f5aa3d07731c2172aae928a9e82892e4d0cbc8808f8ea460a7bc866eb45df7
|
