Trailer Faces HQ Dataset
The Trailer Faces HQ Dataset includes 186,553 high-resolution face images from 15,379 movie trailers, addressing diversity in facial expressions and available under a Creative Commons BY-SA license.
Read original article
The Trailer Faces HQ Dataset, created by Justin Pinkney, consists of 186,553 high-resolution face images sourced from movie trailers. This dataset was developed to address the limitations of existing datasets, such as FFHQ, which lacked diversity in facial expressions. Pinkney collected images from 15,379 trailers available on the Apple Movie Trailers website, resulting in approximately 2 TB of video data. The face detection process utilized a pre-trained Yolov5-face model, and a deduplication method was implemented to ensure only the sharpest images were retained. The final dataset underwent quality filtering to exclude undesirable images, and face identity information was extracted using a model from InsightFace_Pytorch. This allows users to access images of the same individual across different poses and settings. The dataset is available under a Creative Commons BY-SA license, encouraging users to share their applications of the data.
- The dataset contains 186,553 high-resolution face images from movie trailers.
- It was created to provide a diverse range of facial expressions, unlike previous datasets.
- Images were sourced from 15,379 trailers, totaling around 2 TB of video.
- A pre-trained Yolov5-face model was used for face detection and deduplication.
- The dataset is released under a Creative Commons BY-SA license for public use.
Related
Trailer (As Opposite to HTTP Header)
The Trailer response header in HTTP allows senders to add extra fields at the end of chunked messages for metadata like integrity checks. TE header must be set to "trailers" to enable this feature, enhancing data transmission security.
HuggingFace releases support for tool-use and RAG models
The GitHub repository of Hugging Face Transformers provides details on a versatile library for NLP, computer vision, and audio tasks. Users can access it for learning and implementation. For more information, inquire within.
NYPD Coppelgänger: Exploring Cop Data
Sam Lavigne's Coppelgänger tool uses machine learning to match users' faces with NYPD officers, highlighting facial recognition technology's accessibility and raising concerns about its implications for law enforcement and privacy.
Aryn/deformable-detr-DocLayNet – open-source Layout Model
The Deformable DETR model, trained on DocLayNet, achieves 57.1 mAP for object detection using a transformer architecture. It is available on Hugging Face and has been downloaded 108,960 times recently.
Show HN: Hotshot – 4 Person Team Builds a State of the Art Video Model
A four-person team developed Hotshot, a text-to-video model generating 10-second videos at 720p, achieving 70% user preference. The project faced significant data and infrastructure challenges over four months.
2 commentshttps://psych-neuro.com/2016/04/04/hollywood-beauty-does-fam...
>When Schmid compared these aspects of celebrity faces to those of non-celebrities, she found that the average attractiveness score out of 10 for non-celebrities was between 4-5 while celebrities tend to score at a minimum of 6. According to her algorithm, celebrities are significantly more attractive than non-celebrities with top scorers including Brad Pitt, George Clooney, Kate Upton, and Miley Cyrus.
How can this be legal? All imagery is taken from (usually) non-free movie trailers.
Related
Trailer (As Opposite to HTTP Header)
The Trailer response header in HTTP allows senders to add extra fields at the end of chunked messages for metadata like integrity checks. TE header must be set to "trailers" to enable this feature, enhancing data transmission security.
HuggingFace releases support for tool-use and RAG models
The GitHub repository of Hugging Face Transformers provides details on a versatile library for NLP, computer vision, and audio tasks. Users can access it for learning and implementation. For more information, inquire within.
NYPD Coppelgänger: Exploring Cop Data
Sam Lavigne's Coppelgänger tool uses machine learning to match users' faces with NYPD officers, highlighting facial recognition technology's accessibility and raising concerns about its implications for law enforcement and privacy.
Aryn/deformable-detr-DocLayNet – open-source Layout Model
The Deformable DETR model, trained on DocLayNet, achieves 57.1 mAP for object detection using a transformer architecture. It is available on Hugging Face and has been downloaded 108,960 times recently.
Show HN: Hotshot – 4 Person Team Builds a State of the Art Video Model
A four-person team developed Hotshot, a text-to-video model generating 10-second videos at 720p, achieving 70% user preference. The project faced significant data and infrastructure challenges over four months.