Dlib-ml: A Machine Learning Toolkit

By Davis E. King. 2009

3,585 citations recorded by the source corpus.

Subfields: Artificial Intelligence, ML Systems & Hardware, Signal Processing

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In ML Systems & Hardware

In Signal Processing

Citations in the corpus

A Hybrid Model for Identity Obfuscation by Face Replacement (2018, methodology): “= X66 i=1 jjl i u x(p)jj2 2 : (2) l i is the ith landmark’s image position and xis the index of the corresponding landmark vertex on the face mesh. Image landmarks are computed using the dlib toolkit [33]. E photo(p) is a photometric alignment term which measures the pervertex appearance dierence be”
Diversity in Faces (2019, methodology): “Then each detected face as in (a) was processed using DLIB [62] to extract pose and landmark points as shown in (b) and subsequently assessed based on the width and height of the face region.”
VoxCeleb: A Large-Scale Speaker Identification Dataset (2017, methodology): “The HOG-based face detector [32] is used to detect the faces in every frame of the video.”
Hierarchical Deep Temporal Models for Group Activity Recognition (2016, methodology): “For tracking subjects in a scene, we used the tracker by Danelljan et al. [61], implemented in the Dlib library [62].”
On the "steerability" of generative adversarial networks (2019, methodology): “We use the dlib face detector (King, 2009) for bounding box coordinates.”
LipNet: End-to-End Sentence-level Lipreading (2016, methodology): “The videos were processed with DLib face detector (King, 2009) and the iBug face shape predictor with 68 landmarks (Sagonas et al., 2013).”
Large Scale 3D Morphable Models (2017, methodology): “…so we are able to use a HOG active appearance model, a state-of-the-art image landmark localization technique (Antonakos et al. 2014), initialized from a state-of-the-art face detector (King 2009, Alabort-i Medina et al. 2014), in order to robustly locate a set of 68 sparse annotations in the view.”
M2TR: Multi-modal Multi-scale Transformers for Deepfake Detection (2021, methodology): “Implementation Details For all real and forgery images, we use dlib [27] to crop the face regions as inputs with a size of 320 × 320.”
Live Face De-Identification in Video (2019, positive): “The process during test time is similar to the steps taken in the face swapping literature and involves the following steps: (a) A square bounding box is extracted using the ’dlib’ [21] face detector.”
Voxceleb: Large-scale speaker verification in the wild (2020, methodology): “This detector is a distinct improvement over (King, 2009), allowing the detection of faces in pro ﬁ le and extreme poses.”
OpenFace: An open source facial behavior analysis toolkit (2016, methodology): “To initialize our CLNF model we use the face detector found in the dlib library [33, 34].”
Out of Time: Automated Lip Sync in the Wild (2016, methodology): “The HOG-based face detection method of [12] is then performed on every frame, and the face detections are grouped across frames using a KLT tracker [25].”
Toward More Realistic Face Recognition Evaluation Protocols for the YouTube Faces Database (2018, methodology): “The other is the model provided in Dlib open source machine learning library [7], which is a ResNet network with 27 convolutional layers, inspired on the ResNet-34 network from [5].”
Deepfacelab: Integrated, flexible and extensible face-swapping framework (2020, methodology): “We use the default face veriﬁcation method of DLIB [12] for the comparison of identities.”
Drag Your GAN: Interactive Point-based Manipulation on the Generative Image Manifold (2023, methodology): “Since face landmark detection is very reliable using an off-the-shelf tool [King 2009], we use its prediction as ground truth landmarks.”
A General-Purpose Machine Learning Framework for Predicting Properties of Inorganic Materials (2016, background): “, can be run on available hardware in a few hours) and has low error rates in cross-validation—a process that is simplified by the availability of well-documented libraries of machine learning algorithms.(44,45) We often find that ensembles of decision trees (e.”