SignVer [Alpha]: A library for Automatic Offline Signature Verification

SignVer is an experimental library (not under active development) released as part of work on deep learning for offline signature verification. Read more here.

SignVer applies modern deep learning techniques in addressing the task of offline signature verification -
given a pair (or pairs of) signatures, determine if they are produced by the same user (genuine signatures) or different users (potential forgeries). SignVer addresses this task by providing a set of modules that address subtasks required to implement signature verification in real world environments.

Signver Library Modules

Detector

Returns a list of bounding boxes where signatures are located in an image.

from signver.detector import Detector
detector = Detector()
detector.load(detector_model_path)
boxes, scores, classes, detections = detector.detect(img_tensor) 
plot_np_array(annotated_image, plot_title="Document and Extracted Signatures")

Cleaner

Returns a list of cleaned signature images (removal of background lines and text), given a list of signature images

# Get image crops
signatures = get_image_crops(img_tensor, boxes, scores,  threshold = 0.22 )
cleaned_sigs = cleaner.clean(np.array(signatures))

Extractor

Returns a list of vector representations, given a list of image tensors/np arrays

from signver.extractor import MetricExtractor
extractor = MetricExtractor() 
extractor.load(extractor_model_path)
features = extractor.extract(signature_list)

Matcher

Returns a distance measure given a pair of signatures

from signver.matcher import Matcher
matcher = Matcher()
matcher.cosine_distance(feat1,feat2)    # 0.5
matcher.verify(feat1, feat2)    # False