Skip to main content

out-of-the-box computer vision

Project description

Huasca

Computer vision models OOB (out-of-the-bottle).
                 __
                [__]
   ___        .+'. '+.
   )_(       /:;/ _.+'\
   + +       +:._   .++
 .+'+'+.  _  |:._     |
/+::_..+_[_]_+:._CV   |
)_     /_   _\:._     |
+;:    )_``'_(:._     +
+;::+..+;:.._++.____.+'
`+.._..`+...+'

Huasca enables prototyping by prioritizing generalization and rapid development over accuracy.

Step into the cellar and select a bottle of computer visions.
  • Face detection & localization
  • Face classification
    • age
    • gender
  • Object detection & localization
  • Object tracking
  • Object classification w/o localization

Face and object localization include convenient cropping and annotation methods to feed classifiers.

Roadmap

  • v0.3.0 - reduce and combine models to save space
  • v0.4.x - add style transfer
  • v0.4.x - face recognition

Examples

Detection

Detection results include:

  • boxes: Boxes follow PIL format of (left, upper, right, lower)
    • top-left corner is (0,0) and offsets go down/right from there (physics indexing)
  • scores: confidence score for each detected object
  • labels: label description of the object e.g. ['dog','person']
  • portraits: cropped objects from base image (PIL.Image format)
  • base_image: the source image (PIL.Image format)
  • annotated: the source image with objects annotated (PIL.Image format)

Face & Object Detection

# Get a PIL image from somewhere:
image = ...

# Use PIL image as input:
import huasca

faces   = huasca.detect.faces(image)
objects = huasca.detect.objects(image)

# Display the first face
faces.portraits[0].show()

# Check classes
print(objects.labels)

# Retrieve annotated & labeled version of either
faces.annotated.show()
objects.annotated.show()

Face Demographics

# Get a PIL image of a face from face detector:
face = faces.portraits[0]

gender,age = huasca.classify.demographics(face)

Object Tracking

import huasca

data = json.load(json_data)
object_log = huasca.object_tracking.track_objects(data)
output_json = [obj.to_json() for obj in object_log]

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

huasca-0.2.1.tar.gz (45.2 MB view hashes)

Uploaded source

Built Distribution

huasca-0.2.1-py3-none-any.whl (45.2 MB view hashes)

Uploaded py3

Supported by

AWS AWS Cloud computing Datadog Datadog Monitoring Facebook / Instagram Facebook / Instagram PSF Sponsor Fastly Fastly CDN Google Google Object Storage and Download Analytics Huawei Huawei PSF Sponsor Microsoft Microsoft PSF Sponsor NVIDIA NVIDIA PSF Sponsor Pingdom Pingdom Monitoring Salesforce Salesforce PSF Sponsor Sentry Sentry Error logging StatusPage StatusPage Status page