Skip to main content

DeepFake Technologies : Should we embrace ?

 What is Deepfake

Deepfake is a technical way of generating not genuine videos, audio and visual images based on real data inputs. This is commonly archived using deep learning and machine learning technologies. Deep learning technology uses high-power computers to correct their self by its own without any human intervention.

There is quite several deepfake software available in the market. Some of them are

  • FaceApp [https://www.faceapp.com/]
  • DeepFace Lab [https://github.com/iperov/DeepFaceLab]

below shows some popular deepfake creations that we have seen in the past.


The video was first shared by the Dutch deepfake YouTube Channel.


How this is been created

An ordinary way of archiving this is to use deep learning AI programs to identify patterns and similarities of real video, audio and images and change the reality to map the target outcome. These programs use quite several genuine inputs and manipulate the real data to match the target output.

Generative Adversarial Networks (GAN) AI programs are also used for generating deepfake assets. These programs basically contain two components, Generator and Discriminator. The generator creates the assets and Discriminator keeps on evaluating generated assets to identify whether it is fake or real based on real data and provides the feedback to Generator. Deepfake assets will go through rounds of evaluations before producing the final output.

Business use cases

Technology has been used for entertainment, spreading false news and many other negative aspects. However below shows current and future aspects of the technology for Business use cases.

  • Media News - Media News anchors can be introduced to the technology. Breaking news, and generally whether forecast can be easily imitated.
  • Introduces real-life characters to gaming and augmented reality applications.
  • Generating Audiobooks and vocal content based on a real person without actually performing.
  • Use celebrities or VIPs for creating awareness content.
  • Virtual Class Rooms and Virtual Sessions.
  • First line interaction for Virtual agents and support portals and then pass to the real agent.
  • Realistic IVR (Interactive voice response) agents.

Keep in Mind

In all cases, there should be policies, regulations and processes to capture and use of real person’s identity. Technology can be easily used to bring social and cultural damage. 


#technologyinnovation #deepfake #machinelearning #ai #technologytrends


Labels:

Comments

Post a Comment

Popular posts from this blog

How to Link Two List and Create a combine view in the SharePoint 2010

In this way you can join multiple list together and can create a combine view. for an example assume that you are managing a list for some events. And also you are having a participants in separate list. Thus you want to join the two list and create a composite view. SharePoint 2010 allows you to create this kind of view using  Linked Data Source. in this approach you can create your own custom list in the SharePoint. How to Create a Linked Data source Go to SharePoint designer and  go to the link called data sources. And Click the Linked Data Source button in the ribbon. Then SharePoint designer will prompt following kind of a dialog. In there add two list, that you wan to linked together. I'm adding airline schedule and booking list. those are the two list that i wan to merge. then click next. And it will guide you another screen. it will ask you to select either Merge Merge use to combine list which are having same columns definition. for and example we can say th
32 comments
Read more

Motion Eye Docker compose File

Docker compose files are comes in handy when considering container orchestration. Below example shows my docker compose files and folder structure. ---- Your Folder (motioneye)   -- etc   -- lib   -- docker-compose.yaml You can run the docker compose file using docker-compose -d , and etc and lib folder will be automatically populated in the initiation. --- Below shows the content of the docker-compose.yaml file. version: '3' services:   nodered:    image: "ccrisan/motioneye:master-amd64"    container_name: motioneye    restart: always    user: root    ports:      - 8765:8765    volumes:      - "/etc/localtime:/etc/localtime:ro"      - "./etc:/etc/motioneye"      - "./lib:/var/lib/motioneye"
Post a Comment
Read more

How to get Username , UserID in CAML Query

If you are want to get the userID you can simple use following code. <Where>    <Eq>       <FieldRef Name='userfieldname' />       <Value Type='Integer'>            <UserID Type='Integer' />       </Value>    </Eq> </Where> in here you should declare the UserID variable before it uses. normally If you make a user filter in the SharePoint designer it will automatically create the parameter in parameter binding section in the web part. <ParameterBinding Name="UserID" Location="CAMLVariable" DefaultValue="CurrentUserName"/> But assume you want to filter using a user name and the field is not a persongroup feild and it is just a text field. then you can use following query to archive it. <Where>    <Eq>       <FieldRef Name='userfieldname' />       <Value Type=’Text’>            <UserID Type=’Text’/>       </Value>    </Eq> </Wh
9 comments
Read more