Ngene

Anomaly Detection in Simple, Robust and Fast way Training Time: 30 sec Inference Time: 6ms Accuracy(AUROC): >0.95 for more than 70 datasets. No of Training Images; ~100s No of Annotations: 0 With help of our #DeepLTK (Deep Learning) and #CuLab (GPU Acceleration) Toolkits for #LabVIEW we created a very simple, accurate and extremely fast framework for #unsupervised #anomalydetection. For more details, please check: Blog post - https://lnkd.in/dX_riSGJ Example on GitHub - https://lnkd.in/eSjSNgAm https://lnkd.in/esdgi72u

DeepLTK Tutorial: Visual Anomaly Detection in #LabVIEW In this blog post we present a method for #anomalydetection which has several advantages: 1. Unsupervised learning: No dataset annotations are required. 2. Only “good” samples needed for training: Anomalous samples are unnecessary, which is especially beneficial since collecting such data can be challenging. 3. Small dataset requirement: This method can achieve near-perfect accuracy with as few as 60 training samples. 4. High speed: Training typically completes in under a minute, with inference times around 10ms per image. Read more in: https://lnkd.in/ejCWz9VX

[Update] I apologize for my previous post, where I mistakenly shared an old video that lacked some features and was executed on different devices. Here's the correct demo, showcasing: 1. Display of full spectrum of the incoming signal at 40MHz. 2. Dynamic update of waterfall spectrogram. 3. Exponential averaging of the spectrums. 4. Performance metrics for the RTX-2080Ti and Core i7-8700k. This demo clearly shows the GPU outperforming the CPU by 110x. Again, a huge thanks to Norman Kirchner Jr. for his invaluable assistance in developing this demo. Explore our(Ngene) CuLab toolkit for developing #GPU accelerated applications in #LabVIEW: https://lnkd.in/eR7eGuWv

New Update! CuLab - GPU Accelerated by Ngene 📢 This is a major update which brings lots of new functionalities and improvements. To check all the new features, extended functionalities, optimizations and bugs fix go to vipm website: https://lnkd.in/gW6MnxYv Kudos to all contributors!