dcn01

-

收集有关Android自动化测试的框架使用说明，如uiautomator、monkeyrunner

--1

定位综述 目前用于人脸定位的有两种主流方法，一种是基于级联形状回归模型，另一种就是基于深度学习的方法。级联形状回归方法就是使用回归模型，直接学习从人脸表征到人脸形状的映射函数，进而建立从表观到形状的对应关系。这种方法的学习依赖于训练集的选取，如果训练集中包含了复杂的姿态变化，学习到的函数测试性能就会比较好。现有很多基于回归的方法，其中比较突出的有颜水成的DCR（Deep Cascaded Regression）、ESR（Explicitly Shape Regression）方法、LBF（Local Binary Features）方法以及SDM（Supervised Decent Method）方法。但是，上述方法也存在一定的问题，例如，基于提取到的局部图像特征来定位时很难找到具有全局信息的关键点。基于点分布模型的方法和基于形状回归的方法都很依赖于初始值，其中，初始值通常由训练集的平均形状来给定，如果初始值或者设定的平均形状远远偏离目标位置，很难收敛到正确位置。比如，训练集中大部分都是正脸，那么对于测试集中大偏转角度人脸图像的定位就比较困难。 于是引入深度学习来解决较大面部偏转姿态的回归问题，深度学习的最大的优点是有强大的表达能力，可以自学习图像的特征，不需要人为的寻找特征。目前用的基本网络有（1）级联方式：由粗到精一级一级的优化前一步得到的形状，比如香港中文大学汤晓鸥老师的研究团队采用深度卷积神经网络的3个级联结构，逐步细化特征点位置。（2）沙漏网络：密集堆叠连接的U-Nets来进行人脸关键点定位，如CU-Nets通过卷积-反卷积，下采样与上采样网络，跨越不同U-Nets进行全局梯度传播，融合多尺度特征，不断迭代优化关键点坐标。但是深度学习方法的缺点是模型过于复杂，参数非常多，耗时比较长。 深度学习的国内外方法优化：（1）最早的有将人脸进行五官分区域定位回归，但很容易使最终结果陷入局部最优值而不是全局最优。（2）heatmap热图的方法，对数据标签进行处理生成高斯热图，从而更好的回归关键点位置。（3）风格聚合方法，对数据集进行处理生成不同风格的图片，以应对复杂的真实环境如光照强弱。（4）基于边缘感知的人脸关键点检测算法，首先通过消息传递并结合对抗学习得到高精度的边缘线检测结果，再将边缘线信息融合到关键点检测中，来提升算法在大侧脸、夸张表情、遮挡、模糊等极端情况下的鲁棒性。（5）3D人脸模型方法，一种方式是结合3D人脸姿态估计与投影来确定特征点初始位置，然后使用经典的回归树集成ERT方法来更好的进行位置回归；另一种是直接从单幅人脸图像中同时回归出3D人脸结构和密集对齐点；该类方法通过回归位置图，来获得3D几何以及语义信息。（6）人脸特征点检测与形状拟合方法，基于“特征提取”+“回归坐标”的方法，把形状拟合的坐标回归问题，转化为坐标PCA压缩后系数与形状整体仿射变换系数的回归问题（相当于将传统的SDM算法用于深度学习算法当中）。 目前定位方法的改进是： 1.基于深度学习的方法：改进如下： （1）受人体关键点定位 难例挖掘方法的启发，在全局的globalnet之后继续对损失较大的点继续进行微调。 （2）并行集成的方式同时训练两个网络globalnet、refinenet，测试时，第一个网络输出的l2 loss 取top k loss作为refinenet的辅助损失继续微调回归关键点位置。（在300w 有缺失块的数据集上最终的测试结果，平均定位误差为5.9%） 300-W数据库： 68点定位库，包含4个数据集，比如AFW，LFPW，HELLE，还包含135张IBUG里面的图片，IBUG图片对于定位来说都是非常具有挑战性的，不管是旋转角度还是遮挡程度都是非常大的。我们用HELEN的2000张，LFPW的811张，AFW的337张，总共3148张图片作为训练集。用HELEN的554张，IBUG的135张图片，总共689张图片作为测试集。一般测试集分为3个标准：简单，有挑战，全集。简单的测试集就是前面554张图片，有挑战的就是后面那135张图片，全集即689张图片。对三个测试集分别测试，得出测试结果，并与其它方法进行比较。测试标准为定位的平均误差，是被双瞳孔距离归一化后的，平均误差越低代表这一算法越好。

ADBKeyBoard

Android Virtual Keyboard Input via ADB

Android-Architecture-Component

A simple app that uses Android Architecture Components along with dagger2, rxAndroid, rxKotlin, retrofit

Android-CleanArchitecture

This is a sample app that is part of a series of blog posts I have written about how to architect an android application using Uncle Bob's clean architecture approach.

AndroidDropFrameAnalysis

py写的一个评测流畅度的工具,每隔1.2秒执行一下adb shell dumpsys gfxinfo com.miui.xxxx framestats,同时uiautomator操作app,并且对结果去重,最后生成html柱状图展示9个阶段的帧耗时,可以展示几千帧不卡.

DiskCacheManager

Android文件缓存管理： 1、同步读写/异步读写：提供简单易用的api，可以同步读写文件、异步读写文件 2、异步读取并回调到主线程：提供SyncCallback，该回调函数，在异步读取文件后会回调的主线程 3、自动从最老的文件开始清理超过缓存大小限制的文件： CacheManager.getInstance(applicationContext).getDiskCache("dir").setAutoClearEnable(true); 4、线程安全 文件写入有加读写锁，每个文件一把锁，保证线程安全，锁放在对象池中，高效高性能 5、直接读写对象：提供了api，直接读写对象即可。

KotlinDesignPatterns

Kotlin设计模式

Minimal-Todo

Material To-Do App

v2rayN

dcn01/ai_fast_handbook

dcn01/ARES

Black-box tool that uses Deep Reinforcement Learning to test and explore Android applications

dcn01/audio_datasets

Repo containing audio datasets such as GenSpeech and TCD-VoIP. These datasets are commonly used for testing speech and audio quality metrics such as ViSQOL, POLQA, PESQ etc.

dcn01/Cactus-1

用于Android实现保活的策略库

dcn01/DeepHashing

Image hashing is the one-way process of converting an image into a binary hash such that similar images have similar hashes. This has promising applications in speeding up approximate nearest neighbor search when trying to retrieve similar images from a database as well as in security for verifying an image hasn’t been perceptually modified. Utilizing deep learning, we implement a model that learns these binary hashes under three primary constraints. First, we minimize the loss in information between the continuous model output and the quantized binary hash. Second, we make sure the binary values are distributed evenly on each bit. Third, we ensure different bits are as independent as possible through a relaxed orthogonality constraint on each fully connected layer of the model. In addition, we implement a variant of the same model that takes advantage of training data labeled for classification tasks in order to generate hashes that are near one another for images of the same class and far away for images of different classes. We evaulate the supervised and unsupervised variants of this model on the MNIST and CIFAR-10 datasets, as was done in the original paper, as well as a recent malaria diagnosis dataset from the NLM.

dcn01/duplicate-images

A script to find and delete duplicate images using pHash.

dcn01/duplicate_image_finder

find similar or duplicate images

dcn01/ffmpeg-quality-metrics

Calculate quality metrics with FFmpeg (SSIM, PSNR, VMAF, VIF)

dcn01/future-image-similarity

dcn01/Generative-model-most-similar-images

Function that output the 5 most similar training (real) images to the given generated (fake) images.

dcn01/HsexPicComTools

视频封面查找，图片相似度对比

dcn01/image-search-engine

Web app to search similar images

dcn01/imgsmlr-server

Image search service. image search by image. 图片搜索服务. 相似图搜索

dcn01/IPI-FinalProject

Implementation of an image processing algorithm inspired by Kevin Dade Toonify Paper. The main proposal of the project is to achieve a friendly photo filter that transform every photo to a similar cartoon picture up to 5 seconds using Python.

dcn01/JavaBooks

📚Java程序员必读书单整理，附下载地址，助力每一个Java程序员构建属于自己的知识体系。包括但不限于Java、设计模式、计算机网络、操作系统、数据库、数据结构与算法、大数据、架构、面试等等。

dcn01/lipsync

Lip Synchronization (Wav2Lip).

dcn01/lipsync-1

dcn01/LipSync-NEGU93

LipSync by: José Agustín Barrachina & Matias Dwek

dcn01/Media-Displayer

A simple desktop application that allows the user to apply transformative and contrast enhancement techniques onto a specified image. Similar to a tool like Photoshop, the user is able to load either a color or greyscale image onto the main display and, via a panel of buttons, layer transformations onto the original picture.

dcn01/MLAlgorithms

Minimal and clean examples of machine learning algorithms implementations

dcn01/programming-tutorials

dcn01/pypicframe

Smart picture frame software for the Raspberry Pi and similar SBCs

dcn01/RetinaNet-PyTorch

A RetinaNet Pytorch Implementation on remote sensing images and has the similar mAP result with RetinaNet in MMdetection.

dcn01/reverse_image_search

Based on the predicted result of VGG16 model and reordering probably similar images via pre-trained Xception or ResNet model.

dcn01/similar_image_websearch

Search near duplicate or similar images in your gallery by uploading an image. Algorithms used (perceptual hash, mean squared error, structural similarity index)

dcn01/similarity_search

Reverse image search, similarity models, and image retrieval

dcn01/sudoku-lens

This project is the implementation of the sudoku solver using computer vision & backtracking algorithm

dcn01/text_renderer

Generate text images for training deep learning ocr model

dcn01/TextRecognitionDataGenerator

A synthetic data generator for text recognition

dcn01/VisemeNet

This is the visemeNet for lipSync