Fingerprints are widely recognized as one of the most unique and reliable characteristics human identity. Most modern fingerprint authentication systems rely on contact-based fingerprints, which require use scanners or sensors for capturing fingerprints during process. Various types sensors, such optical, capacitive, ultrasonic employ distinct techniques to gather analyze data. This dependency specific hardware creates a barrier challenge broader adoption based biometric systems. limitation...

Supermarkets need to ensure clean and safe environments for both shoppers employees. Slips, trips, falls can result in injuries that have a physical as well financial cost. Timely detection of hazardous conditions such spilled liquids or fallen items on supermarket floors reduce the chances serious injuries. This paper presents EdgeLite, novel, lightweight deep learning model easy deployment inference resource-constrained devices. We describe use EdgeLite two edge devices detecting floor...

The use of transfer learning with deep neural networks has increasingly become widespread for deploying well-tested computer vision systems to newer domains, especially those limited datasets. We describe a case domain data-starved regime, having fewer than 100 labeled target samples. evaluate the effectiveness convolutional feature extraction and fine-tuning overparameterized models respect size training data, as well their generalization performance on data covariate shift, or...

Supermarkets need to implement safety measures create a safe environment for shoppers and employees. Many of these injuries, such as falls, are caused by lack precautions. Such incidents preventable timely detection hazardous conditions undesirable objects on supermarket floors. In this paper, we describe EdgeLite, new lightweight deep learning model specifically designed local fast inference edge devices which have limited memory compute power. We show how EdgeLite was deployed three...

Resistive computing systems (RCSs) are projected to be leveraged as inference engines for Deep Neural Networks (DNNs). Unfortunately, limited device yield due immature fabrication processes may severely degrade the DNN's classification accuracy. The arising solution is leverage resilient-aware data layout organization techniques mask defects using neural network weights. However, current too slow practical real-world applications. In this paper, we propose a framework fast enable large DNNs...

Many fingerprint recognition systems capture four fingerprints in one image. In such systems, the processing pipeline must first segment each four-fingerprint slap into individual fingerprints. Note that most of current segmentation algorithms have been designed and evaluated using only adult datasets. this work, we developed a human-annotated in-house dataset 15790 slaps which 9084 are samples 6706 drawn from children ages 4 to 12. Subsequently, is used evaluate matching performance NFSEG,...

Fingerprint-based identification systems achieve higher accuracy when a slap containing multiple fingerprints of subject is used instead single fingerprint. However, segmenting or auto-localizing all in image challenging task due to the different orientations fingerprints, noisy backgrounds, and smaller size fingertip components. The presence images real-world dataset where one more are rotated makes it for biometric recognition system localize label automatically. Improper fingerprint...

The integration of artificial intelligence capabilities into modern software systems is increasingly being simplified through the use cloud-based machine learning services and representational state transfer architecture design. However, insufficient information regarding underlying model provenance lack control over evolution serve as an impediment to more widespread adoption these in many operational environments which have strict security requirements. Furthermore, tools such TensorFlow...

Many fingerprint recognition systems capture four fingerprints in one image. In such systems, the processing pipeline must first segment each four-fingerprint slap into individual fingerprints. Note that most of current segmentation algorithms have been designed and evaluated using only adult datasets. this work, we developed a human-annotated in-house dataset 15790 slaps which 9084 are samples 6706 drawn from children ages 4 to 12. Subsequently, is used evaluate matching performance NFSEG,...