-
tutorial-face-mask-detection
In this project, we develop a pipeline to detect unmasked faces in images. This can, for example, be used to alert people that do not wear a mask when entering a building.
-
InfluxDB
Power Real-Time Data Analytics at Scale. Get real-time insights from all types of time series data with InfluxDB. Ingest, query, and analyze billions of data points in real-time with unbounded cardinality.
-
DSFD-Pytorch-Inference
A High-Performance Pytorch Implementation of face detection models, including RetinaFace and DSFD
-
face-mask-detection
Discontinued In this project, we develop a pipeline to detect unmasked faces in images. This can, for example, be used to alert people that do not wear a mask when entering a building. [Moved to: https://github.com/datarootsio/tutorial-face-mask-detection] (by datarootsio)
Our validation and testing data consists of images of people with and without masks that we collected from various sources that provide images with permissive licences (e.g. pexels.com, unsplash.com). We have manually annotated all faces in the collected images, and labeled them as being masked or not (using the makesense.ai annotation tool). We collected 273 images which contain 524 faces (246 masked and 278 non-masked). The images are split 50/50 over the validation set and test set. An overview of the collected data and corresponding URLs and ground truth annotations can be found in test_validation_metadata.csv.
This strategy is based on the description that you can find in the prajnasb/observations repository. We apply 13 masks with different shapes and colors to generate training data, which you can find in data/mask-templates. Below you can see an example of a mask being artificially applied.
We used the RetinaFace face detector to extract faces as it is the state-of-the-art in face localisation in the wild, and works in real-time on a single CPU core (Deng et al.). We used the implementation and pre-trained model available at the RetinaFace repository. Note that we used the RetinaNetMobileNetV1 model, which is much faster than RetinaNetResNet50 and DSFDDetector.
git clone https://github.com/datarootsio/face-mask-detection.git cd face-mask-detection mkdir data tar -xvf data.tar.gz -C ./data pip install -r requirements.txt
Related posts
-
ESpeak-ng: speech synthesizer with more than one hundred languages and accents
-
2024 Verizon Data Breach Investigation Report [pdf]
-
Impact of Input Length on the Reasoning Performance of Large Language Models
-
Kolmogorov-Arnold Networks
-
Quick tip: Write numpy arrays directly to the SingleStore VECTOR data type