Do you want to move your image/video editing software into the 21st century?
If so Mut1ny's Face processor might be something you should consider. Mut1ny's face processor is a deep learning based facial/upper body image based processing pipeline using Mut1ny's unique expertise, unique assembled training data and constantly evolving and updated deep neural networks for smart facial image processing.
What can Mut1ny's Face processor do?
With our recent upgraded series of deep convolution neural networks. Mut1ny's face processor array of processing networks contains a state of the art segmentation network that allows to extract per-pixel facial area mask from any image content. Separable masks for
- Eye brows
- Other facial hair
facial areas can be generated. The segmentation network is trained on an extended version of our Face/Headsegmentation dataset
But that is not all, additionally Mut1ny's face processor contains trained networks that can also extract per-pixel:
- Ambient occlusion / surface accessibility
- Surface orientation
- 3D Positional data (under development)
- Other lighting information (under development)
with just the single image no additional sensory data, series of images or enriched RGB content is required.
See for yourself a series of test images that have been processed by our trained networks (this is the raw output of the networks without any further post-processing):
|Source||Segmented||Ambient Occlusion||Depth||Normal (lambert shading demo)|
Although not optimized for temporal content here a demonstration on a short sequence
Another demonstration video face processor used on the whole LFW dataset to just separate the face from the background
Faceprocessor LFW demo
How can we help you to take advantage of our face processor and make your product more innovative?
Here at Mut1ny we can offer you different level of access, support and partnership to the face processor
Raw data partner subscription model
Trained network subscription model
Full integration into your product
|You have enough deep and machine learning expertise in your company to train your own networks but are interested in the raw training material then this level of partnership is probably most suitable for you. At this level you simply will receive monthly updates on the raw data that we internally use to train our networks.||Since we continuously train and improve our networks you can benefit from our continuous work and expertise in this area. On this level we provide you with our trained network models in ONNX format that you then can immediately use with the deployment framework/platform of your choice inside your application.. You can not use ONNX in your application no problem tell us what other format you need and we see that we can provide it.||You have absolutely no experience in using deep learning or how to deploy it into your product? Or you do not have the manpower available right now to do this on your own. Or maybe you simply just want things to work without spending much development time of your own? Then this is probably the level of partnership you want to explore. In this case we will provide full support and help you to find the best solution for your target platform and customers. We help you with integrating our deep learning models directly into your application all the way to until the finished integration and beyond.|
Interested to explore any of the options above, please use the contact form below and we get back to you as soon as possible