SlapSeg III Evaluation

These comments provided by Neurotechnology are based on the submission results, reviewed on April 14, 2022.

Neurotechnology's slap fingerprint segmentation algorithm has been judged by NIST as the most accurate among the SlapSeg III 2 inch and 5.5 inch category participants, as well as second most accurate in the 8 inch category.

What is SlapSeg III?

SlapSeg III Evaluation is a NIST-run public test of automated slap fingerprint segmentation algorithms. Output from the tested automated algorithms is compared to output from a human examiner and analyzed for similarity.

See SlapSeg III official page for more information.

Fingerprint segmentation is the act of separating or segmenting an image of the friction ridge structure of the hand into individual images of the upper-most finger joints, known as distal phalanges.

Currently there are 7 companies participating in the evaluation. Our latest submission was Neurotechnology+0008. The submitted algorithm is available in MegaMatcher SDK and VeriFinger SDK.

Neurotechnology evaluation results in all 4 categories of the SlapSeg III are presented below:

• 2 inch are scanned images of inked tenprint card slaps.
• 3 inch are live scan images captured using devices with a scanning surface of 3 inches.
• 5.5 inch are half palm (upper palm) live scan images.
• 8 inch are full palm live scan images.

Segmentation accuracy in this category is measured by percentage that at least 8 fingers were segmented correctly. According to this accuracy measurement our algorithm is the most accurate among the participants. Also our algorithm is the second fastest algorithm in the 2 inch category, while the faster contender showed much worse accuracy. Overall Neurotechnology+0008 showed the best speed and accuracy combination in this category.

Speed and accuracy

Click to zoom

Accuracy by fingers

Click to zoom

Segmentation accuracy in this category is measured by percentage that all 10 fingers were segmented correctly. Our algorithm provides perfect balance between accuracy and speed according to the measurement, as more accurate contenders are much slower, and faster contenders show worse segmentation accuracy.

Speed and accuracy

Click to zoom

Accuracy by fingers

Click to zoom

Segmentation accuracy in this category is measured by percentage that at least 8 fingers were segmented correctly. According to this accuracy measurement our algorithm is the most accurate among the participants. Also our algorithm is the second fastest algorithm in the 5.5 inch category, while the faster contender showed much worse accuracy. Overall Neurotechnology+0008 showed the best speed and accuracy combination in this category.

Speed and accuracy

Click to zoom

Accuracy by fingers

Click to zoom

Segmentation accuracy in this category is measured by percentage that at least 8 fingers were segmented correctly. According to this accuracy measurement our algorithm is the second most accurate among the participants. Also our algorithm is the fastest algorithm in the 8 inch category, while the more accurate contender was much slower. Overall Neurotechnology+0008 showed the best speed and accuracy combination in this category.

Speed and accuracy

Click to zoom

Accuracy by fingers

Click to zoom