Leo WangNLP+CV Helps Achieve Zero-Shot Learning With SOTA PerformanceLeveraging the CV model’s zero-shot ability by introducing text encoders and massive amount of noisily-labeled data. Achieving SOTA result6 min read·Oct 10, 2023----
Leo WangSwin Transformer V1 + V2 — Best Vision Models Are Not CNN-based🔥 Swin Transformer has revolutionized computer vision while can still be general-purpose. V2 does even a better job and beats SOTA…5 min read·Oct 8, 2022----
Leo WangCreate Art Words with Data in Python💡 Break the norm — a fast and creative project using python’s most popular data visualization tool to create creative art words.3 min read·Sep 20, 2022----
Leo WanginTowards AITransUNet — No More CNNs for Image SegmentationRenovating U-Net by combining CNNs and Transformer to achieve SOTA result on image segmentation tasks4 min read·Sep 4, 2022--4--4
Leo WanginTowards AIEfficientNet — An Elegant, Powerful CNN.No Fancy Techniques but worked extremely well.5 min read·Aug 30, 2022----
Leo WangAxial Self-Attention in CNN — Efficient and ElegantBy proposing axial attention, attention-based CNNs can now process images more efficiently yet still retain global receptive field.3 min read·Aug 5, 2022----
Leo WanginTowards AIExplained: Reverse Attention Network (RAN) in Image SegmentationTable Of Contents5 min read·Jul 31, 2022----
Leo WanginTowards AIUNet++ Clearly Explained — A Better Image Segmentation ArchitectureIn this article, we are going to introduce you to UNet++, essentially an upgraded version of UNet. This article is designed to help you…5 min read·Jul 22, 2022--1--1
Leo Wang10 Important Linux Commands For Data ScientistAs a data scientist, you may often need to deal with data or models with extremely large sizes, that can’t be fitted into your computer’s…6 min read·Jul 19, 2022----