Product Description

The facial ambient occlusion estimation neural network, estimates the ambient occlusion or surface accessibility of every pixel in the face. Normally to compute this result one would have to have a corresponding 3D mesh model of the face and use it in combination with sophisticated rendering techniques. Our neural network can estimate the ambient occlusion by just using the single RGB input image. It does not require any additional information.

You have total freedom in how you want to deploy the model. You can put it on a server, public or private cloud. Or alternatively use it directly in your application or on an edge device. This choice is totally up to you. There are no constraints and you do not pay any transaction fees for using the model from Mut1ny

The model is deliverable in these supported neural network formats*

*In your order please use the additional field to indicate which neural network format you would prefer

What is included in the facial ambient occlusion estimation deep neural network?

  • The model in the network framework format of your choice (see above)
  • For ONNX format only:
    • an inference script in Python using ONNXRuntime our recommended ONNX inference framework
    • C# code for using our Model using  ONNXRuntime
    • C++ code for using our Model using ONNXRuntime
    • A deployment script for using our Model in combination with AzureMachineLearning
  • For PyTorch only:
    • an inference script in Python using PyTorch
    • C++ code for using our Model using PyTorch
  • For mxnet only

Why buy a deep neural network model? When you can build and train one yourself? Or get one free from the internet? 

  • Training a segmentation model from scratch using a  large enough dataset takes between 2-3 days even with a high-end equipped GPU. This translates to € 80-150,- public cloud instance spending costs alone.
  • No need for you to spend  development time instead you can fully concentrate on your actual product feature.
  • Getting a dataset like this is very hard.
  • To our knowledge there are not any public ambient occlusion estimation networks out there.

Why go for a subscription instead of single one-off purchase?

  • Subscription allows you take advantage of Mut1ny’s future neural network model improvements. We improve our neural network models on a permanent basis adopting to latest research developments. This mean you benefit from our research and therefore will get better models over time.
  • Subscription allows you take advantage of Mut1ny’s constantly growing training set. We constantly enlarge our training set with new data to cover a wider variety. Also if you are  subscription user and you encounter cases that do not produce a result to our satisfactions you can send them to us. We’ll be including those in our training set.


There are no reviews yet.

Only logged in customers who have purchased this product may leave a review.