文勋喆的博客 | Xunzhe Blog

Posted by vince on November 5, 2021

Semantic Segmentation

"Review of semantic & instance segmentation"

Semantic Segmentation Architectures Until state-of-the-art networks for segmentation: 1. FCN 2. LinkNet 3. UNet 4. SegNet 5. DeepLab Evaluation Metrics Dice coefficient: intersection ove...

Posted by vince on November 15, 2020

Single Stage Object Detection

"Create a custom single-stage detector"

The pipeline to create a custom single stage detector: 1. Single-stage NN architecture As the network pipeline, the feature extractor is going to be the combination of ResNet-18 (pre-trained) and...

Posted by vince on November 2, 2020

OCR Demo: Automatic Number Plate Recognitions

"Roubut OCR pipeline using CRAFT & Tesseract"

This blog is the notbook for OCR fundamental studies, which will include: 1. Graphic Text Recognition: Tesseract, Keras-OCR 2. Text Detection: EAST, CRAFT Realistic scenario: text in imagery 1. Gr...

Posted by vince on September 15, 2020

Using YOLO v3 & v4 to Train Custom Object

"YOLO v3 v4 training on Colab"

In this project, it will go through the pipeline for training object detector using YOLO v3 and v4. Training Pipeline 1. Change Runtime Type Runtime -> Change runtime type -> Select hardwar...

Posted by vince on August 30, 2020

Fundamentals of Object Detection Reviews

"Basic Object Detection Notebooks"

Object Detection Challenges: Intra class variance Pose variation Occlusion / Large search space over multiple: (a) Location, (b) Scale, (c)Aspect ratio; Crowded scenes Traditional Objec...

Posted by vince on August 19, 2020

LibTorch on Local System

"Cmake for compiling LibTorch code"

LibTorch Installation Step 1: Download libtorch.zip file Download libtorch zip file (debug or release) from pytorch.org and select the stable version specific to your OS. If have a GPU, select C...

Posted by vince on July 24, 2020

Accelerated Computing (III)

"Asynchronous Streaming, and Visual Profiling with CUDA C/C++"

The CUDA tookit ships with the Nsight Systems, a powerful GUI application to support the development of accelerated CUDA applications. Nsight Systems generates a graphical timeline of an accelerat...

Posted by vince on July 22, 2020

Accelerated Computing (II)

"Managing Accelerated Application Memory with CUDA Unified Memory and nsys"

The CUDA Best Practices Guide, a highly recommended followup to this and other CUDA fundamentals labs, recommends a design cycle called APOD: Assess, Parallelize, Optimize, Deploy. In short, APOD ...

Posted by vince on July 12, 2020

Accelerated Computing (I)

"Accelerating Applications with CUDA C/C++"

CUDA provides a coding paradigm that extends languages like C, C++, Python, and Fortran, to be capable of running accelerated, massively parallelized code on the performant parallel processors: NV...

Posted by vince on July 10, 2020

Facial Landmarks Detection

"CV applications for facial landmarks"

This repo are implementing facial landmarks detection and its applications. Facial landmarks can be achieved by a variety of techniques such as DNN. This repos implemented landmark detection using ...

Posted by vince on June 20, 2020

Clustering: K-Means, DBSCAN

"Implementation of clustering algorithm"

Machine Learning using K-Means and DBSCAN algorithm A cluster refers to a collection of data points aggregated together because of certain similarities. Load data Load the collected data and visu...

Posted by vince on February 6, 2020

MobileNet v1 & v2

"A lite convolutional neural net model"

MobileNet MobileNet is designed for CNN deployment in embedded devices, such as MCU, Mobile phone. It has proven faster speed on inference with little accuracy drop. (compared with other nets such...

Posted by vince on February 5, 2020

Bagging vs Boosting, and Random Forest

"Ensemble learning techniques: bagging & boosting"

Bagging & Boosting 主要思想就是寻找多个识别率不是很高的弱分类算法比寻找一个识别率很高的强分类算法要容易得多 (三个瓜皮匠, 赛过诸葛亮), AdaBoost 算法的任务就是找几个精确度不是那么高的弱分类算法，让它们一起合作形成一个强分类算法，从而提高准确率。 Bagging (Bootstrap aggregation): Bootstrap就是鞋带，帮助提...

Posted by vince on February 3, 2020

ABOUT ME

Xunzhe Blog

PointCloud Processing

"Deep learning in 3D PointCloud"