Tagged articles
feature extraction
32 articles · Page 1 of 1
This article details how NetEase Cloud Music leverages multimodal large language models to improve music recommendation across daily, personalized, and playlist scenarios by extracting rich audio, text, and visual features, addressing data skew, cold‑start challenges, and achieving measurable gains in user engagement and distribution efficiency.
NetEase Cloud Musicfeature extractionlarge language models
0 likes · 12 min read

This article explains the concepts of data dimensionality reduction and feature extraction in machine learning and demonstrates how to implement them in PHP using the PHP‑ML library, including installation, data preprocessing, PCA-based reduction, and feature extraction with token vectorization and TF‑IDF.
PCAPHP-MLdimensionality reduction
0 likes · 5 min read

This article explains the importance of data dimensionality reduction and feature extraction in machine learning, and provides a step‑by‑step guide with PHP code examples—including library installation, data preprocessing, PCA‑based reduction, and feature selection techniques—demonstrating how to handle large datasets efficiently.
PCAPHPdata preprocessing
0 likes · 6 min read

This article introduces the Python library librosa, outlines its main audio processing features such as loading, visualization, MFCC, pitch detection, chromagram, and rhythm analysis, and provides complete code examples for each operation.
MIRPythonaudio processing
0 likes · 7 min read

This article presents a comprehensive guide to automating audio testing in first‑person shooter games, covering common audio bugs, a step‑by‑step recording workflow, feature extraction techniques, similarity metrics, and practical toolchains such as Soundcard, fmedia, VBCable, and librosa.
audio testingautomationfeature extraction
0 likes · 13 min read

This article reviews the fundamentals, feature extraction methods, evaluation metrics, and common datasets of content‑based image retrieval (CBIR), discussing traditional low‑level features, local descriptors, unsupervised and supervised learning approaches, and recent deep‑learning models for improving retrieval performance.
CBIREvaluation Metricsdatasets
0 likes · 13 min read

This article walks through implementing a reverse image search system, starting with simple pixel‑based geometric distance, then improving accuracy using ResNet‑derived feature embeddings, accelerating queries with K‑means clustering, and finally deploying a Milvus vector database for fast, scalable similarity retrieval.
MilvusResNet50clustering
0 likes · 17 min read

This article explains the concept of image‑based search, why it matters, and provides a step‑by‑step guide to implement image search using Elasticsearch 8.x, feature‑extraction libraries, and the multilingual CLIP‑ViT‑B‑32 model, including code snippets and architecture overview.
clip modeldeep learningfeature extraction
0 likes · 10 min read

This article describes the design, algorithmic improvements, and engineering performance optimizations of Bilibili's massive video duplicate detection (collision) system, covering challenges of low‑edit‑degree reposts, two‑stage retrieval, self‑supervised feature extraction, GPU‑accelerated preprocessing, and the resulting gains in accuracy and throughput.
BilibiliLarge-Scale RetrievalVector Search
0 likes · 17 min read

Bilibili built a massive video‑duplicate detection platform that trains a self‑supervised ResNet‑50 feature extractor, removes black borders, and uses a two‑stage ANN‑plus‑segment‑level matching pipeline accelerated by custom GPU decoding and inference, boosting duplicate rejection 7.5×, recall 3.75×, and cutting manual misses from 65 to 5 per day.
GPU accelerationLarge-Scale Retrievaldeep learning
0 likes · 19 min read

The article introduces PaddleBox and FeaBox, two GPU‑driven frameworks designed for massive sparse DNN training and unified feature extraction, detailing their architecture, performance advantages, hardware‑software co‑design challenges, and successful deployment across Baidu's advertising systems.
FeaBoxGPUPaddleBox
0 likes · 24 min read

The article provides a comprehensive overview of Baidu's Wanxiang system, detailing how it tackles the challenges of massive image and video data processing, feature extraction, cross‑media indexing, and real‑time retrieval to support modern search engine products.
BaiduRich Mediabig data
0 likes · 13 min read

The article outlines the Tencent Advertising Algorithm Competition’s video ad challenge, details the paper submission guidelines, and shares a participant’s step‑by‑step technical approach—including baseline experiments, model re‑implementation with Paddle, multimodal feature extraction, optimizer choices, and future improvement directions—providing practical AI insights for multimedia video classification.
Tencent competitiondeep learningfeature extraction
0 likes · 7 min read

This article explains how to use the Hilbert transform to obtain signal feature values, defines R and J envelope statistics and bispectrum features, and provides Python code that generates signals, adds noise, computes these features, and visualizes their variation with amplitude, frequency, and phase.
Hilbert transformPythonSignal Processing
0 likes · 13 min read

This article reviews common image similarity detection techniques—including hash-based methods (aHash, pHash, dHash), histogram comparison, feature matching with ORB and SIFT/SURF, bag‑of‑words with K‑Means, and CNN‑based VGG16 features—detailing their algorithms, Python implementations, performance characteristics, and practical considerations.
Computer Visiondeep learningfeature extraction
0 likes · 15 min read

Dewu Photo Search evolved from a limited Aliyun‑based prototype to a self‑developed pipeline using EfficientNet detection and 128‑dim embeddings, boosting top‑1 shoe accuracy over 100 % and overall precision by up to 41 %, while reducing latency and improving scalability despite remaining stability challenges.
deep learningfeature extractionimage search
0 likes · 10 min read

This article explains the fundamentals of machine learning by illustrating how humans recognize cats, extracting features, training models with gradient descent, implementing linear regression, and providing practical JavaScript and SQL code examples for building a simple classifier.
Gradient DescentJavaScriptSQL
0 likes · 12 min read

This article walks through the complete engineering workflow for building an image‑search system, covering CNN‑based feature extraction with VGG16, vector normalization, image preprocessing, black‑edge removal, and practical deployment of the Milvus vector database including hardware requirements, capacity planning, collection/partition design, and search result handling.
CNNMilvusPython
0 likes · 11 min read

This article explains the fundamentals of building an image‑by‑image search system, covering image feature extraction methods such as hashing, traditional descriptors, CNN‑based vectors, and the use of vector search engines like Milvus for similarity retrieval.
CNNMilvusVector Search
0 likes · 6 min read

The Beihang University team’s video copyright detection solution samples frames every 200 ms, extracts 512‑dimensional ResNet‑18 features, and uses handcrafted cosine‑similarity matching to identify source videos and plagiarized segments, all while operating on limited hardware without training any models.
algorithm designfeature extractionframe sampling
0 likes · 12 min read

This article details a two‑stage video copyright detection pipeline that builds a frame‑level feature library, uses Hessian‑Affine + SIFT and Fisher Vectors for robust feature extraction, applies weighted bipartite graph matching and longest increasing subsequence localization, and achieves an F1‑score of 0.9086 on the CCF 2019 competition dataset.
AIfeature extractionframe matching
0 likes · 14 min read

This article examines iQIYI's end‑to‑end personalized poster production and distribution system, detailing AI‑driven image cropping, smart frame extraction, feature extraction, multi‑armed bandit ranking, and online experiments that together significantly increased poster click‑through rates on TV and mobile platforms.
AI poster generationPersonalized RecommendationVideo platform
0 likes · 12 min read

This article explains how search engines collect, analyze, and rank web pages by describing the spider system, HTML and layout tree construction, feature extraction, and machine‑learning based classification methods used to understand page content and improve result relevance.
HTML treefeature extractionlayout tree
0 likes · 8 min read

This article describes how 58.com’s commercial engineering team redesigned its real‑time feature‑mining pipeline—replacing a minute‑level Spark Streaming framework with Flink—to achieve sub‑second latency, higher throughput, stronger fault‑tolerance, and end‑to‑end exactly‑once semantics for user‑profile generation in the second‑hand‑car recommendation scenario.
FlinkSparkbig data
0 likes · 14 min read

This article provides a thorough introduction to image‑search technology, covering its general framework, offline and online components, feature‑extraction evolution, retrieval engine structures, and architectural challenges such as dynamic indexing, feature synchronization, and high‑throughput low‑latency serving.
Computer Visionfeature extractionimage search
0 likes · 12 min read

This article summarizes a computer‑vision salon where Dr. Ji Yongnan explains imaging pipelines, traditional feature‑based methods, deep‑learning breakthroughs, Tencent Cloud AI services, real‑world case studies, and answers audience questions about machine‑vision versus computer‑vision and data‑scarcity challenges.
AI applicationsComputer VisionObject Detection
0 likes · 18 min read

This article explains why simple MD5 checks cannot assess image similarity and introduces two major approaches—image hashing and image feature extraction—detailing their algorithms, practical performance, and how to compare images efficiently using Hamming distance and indexing techniques.
Computer Visionfeature extractionhamming distance
0 likes · 7 min read

This article presents a detailed case study of building a purchase‑user prediction model by integrating Convolutional Neural Networks for feature extraction with Long Short‑Term Memory networks for time‑series forecasting, covering background, model structure, data augmentation, experimental results, and business impact.
CNNLSTMModel Evaluation
0 likes · 10 min read

This article presents a technical solution for extracting representative key frames from time‑series screenshots of a mobile app installation process, covering pixel sampling, dimensionality reduction, classic feature extractors (SIFT, HOG, ORB), auto‑encoder based deep learning, and clustering methods such as KMeans and DBSCAN, along with practical results and performance analysis.
Computer VisionHOGORB
0 likes · 5 min read

DeepHash is Meitu’s large‑scale short‑video analysis and retrieval system that converts deep‑learned visual features into compact binary hash codes via a MobileNet‑based CNN and triplet‑loss training, enabling fast, robust similarity search across billions of videos with sub‑second latency and minimal storage.
feature extractionlarge-scalemultimedia retrieval
0 likes · 15 min read

This study compares hash‑based, CNN‑based, and local‑feature methods for client‑side image similarity detection in e‑commerce, showing that while hash methods are fast and CNNs are accurate but costly, the Hessian‑Affine detector combined with SIFT descriptors delivers the optimal balance of computational efficiency, robustness to transformations, and high recall/precision for on‑device duplicate filtering.
CNNHash AlgorithmsSIFT
0 likes · 11 min read

The article discusses the evolution of image search engines, outlines key challenges such as image quality, watermarks, speed, and feature extraction, and explains how the Facenet deep‑learning model with Triplet loss can be used to generate compact image embeddings for efficient similarity search.
Computer Visiondeep learningfacenet
0 likes · 7 min read
