पंजाब यूनिवर्सिटी का AI मॉडल: अब हादसे में खराब चेहरे की दांतों और जबड़ों से होगी पहचान, भारत सरकार से मिला कॉपीराइट

ETVBHARAT

पंजाब यूनिवर्सिटी के साइंटिस्ट विभाग ने "फेशियल रिकग्निशन मॉडल" तैयार किया है, जिससे हादसे के बाद खराब चेहरे की पहचान हो सकेगी.

Transcript

00:00Punjab University has then confirmed that he is the first one in his research.

00:06We are in Punjab University in the Anthropology Department where we are with Professor Kevil Kishan.

00:12The team has then confirmed that they can see the faces of their faces,

00:18which are in their faces, and their faces are in their faces.

00:26So they have also prepared a model that has been done in their research.

00:33Our team and professors are talking to them.

00:36Sir, this model was prepared.

00:39You are working with the first time in the forensic lab.

00:43So this model is the first idea.

00:48When did it come and how did it start?

00:50This idea came to our mind when Damani had a PhD topic.

00:56It came to our mind that why did we work on dental science and teeth and jaws?

01:03When we started working on teeth and jaws,

01:06we thought that the dimensions of teeth and jaws can become very good.

01:11So Damani and my team started learning AI.

01:16We started learning AI.

01:18We started learning AI.

01:20When we started learning AI,

01:23we started learning AI.

01:25Then we started putting it into our teeth.

01:28We started collecting teeth and collecting data.

01:31So we started collecting data.

01:33We started collecting data collection.

01:34They started collecting data from about 800 people.

01:38So this is a very big data collection,

01:40Damani and Nandani,

01:42who did their PhD thesis.

01:44We have developed teeth dimensions.

01:49We created a model of teeth and face dimensions,

01:54and we created machine learning.

01:55We started learning different models.

01:57And the most good model is our 95% accuracy.

02:02It means that when we have to create teeth with a face or jaw,

02:12then the face will be 95% accuracy.

02:15It's not the same ever.

02:17We believe that the accuracy of DNA is 100%.

02:21But this is all morphology.

02:23And in the work of morphology, there is no accuracy in 100%.

02:27But we are close to 100%.

02:29And we can recreate their face.

02:33When there are mass disasters, bomb blasts,

02:38and we don't know the faces,

02:41or the faces are beyond recognition.

02:43In that case, teeth are the hardest part of the body.

02:46So we get to know the remains of teeth.

02:49That is the most important thing.

02:53We want to do that.

02:55And this is our program.

02:57This is our EIE model.

02:59And we have copyrighted by Government of India.

03:03You have received copyright.

03:05And you are saying that 95% of your model is working.

03:09So now, if we take the example of Ahmedabad's case,

03:13which was a plane crash,

03:15do you have transferred to those types of cases?

03:19And you are working on it?

03:21We haven't thought that we will work on such a big level.

03:25But I thought that when we have teeth dimensions,

03:29why can we create face-create them?

03:31We could also create face-create them.

03:33But there is no problem.

03:35Our copyright has come after that.

03:37And we have seen the practicality.

03:41And when we tested it,

03:43we tested some jaws.

03:45So we know that 95% accuracy is 95% accuracy.

03:49So if you understand how to do this model,

03:53then what will it work?

03:55If you want to tell us how to tell us?

03:57Damini, tell us.

03:59Damini, tell us.

04:01Because this whole model is a very important role.

04:05And the rest of the people also have a good job.

04:07So what will this model work?

04:09Can we explain?

04:11Basically, the model we have prepared

04:13is Python, which is a large-level language of a computer.

04:17When we have prepared,

04:19we have trained the teeth measurements before.

04:21We have given the teeth measurements

04:23and teeth measurements.

04:25When it is trained,

04:27we have given the unknown teeth measurements.

04:29And we have given the measurements of 95% accuracy.

04:33So we have compared

04:35the actual measurements

04:37and the predicted measurements.

04:39Which is very close.

04:41Which is very close.

04:43We know that at least this close face shape

04:45which is the front face of the person.

04:47Which is the most important for identification.

04:49Which is the most important for identification.

04:51So we also have to know

04:52which measurements are the teeth

04:53which is the most important prediction.

04:55So like our canines,

04:57our molars,

04:59which is the most important for identification.

05:01So this model works.

05:03Sir has told us that

05:05you have taken samples from 800 to 840.

05:09You have taken samples from 800 to 840.

05:11You have taken samples from a report

05:13and then researched it.

05:15So where did you get these samples from?

05:17How did you collect these samples?

05:19How did you collect this data?

05:21And how many times did you work?

05:23So the sample collection is the most difficult part of the PhD.

05:27So as Sir's guidance,

05:29Sir has told us,

05:31I have collected data from Himachal

05:33and Dr. Nandini Chitara

05:35collected data from Haryana.

05:37So we have a diverse data

05:39like 800 to 800 to 850.

05:41So we collected all the data

05:43and then we collected it in Excel sheet.

05:45and then we collected all the data.

05:47Now if we talk about the future,

05:49what is your plan in the future?

05:51Is it a whole team working with you?

05:53Or is it the forensic department

05:55who is working in these cases?

05:57Will you help them?

05:59Or will there be some transfer cases here?

06:01The cases will not be transferred here, ma'am.

06:03Because the forensic department

06:05can be able to work with them.

06:07But it is that they can use the model

06:09but they can use the model.

06:11They can use the model.

06:13But it is that they can use the model.

06:15And we also want,

06:16we also want our model to be used.

06:18And they also know that

06:20how much accuracy is coming from the model.

06:22So we can do some improvement

06:24We can do 97% accuracy.

06:26And if we have more data,

06:28we can do that.

06:30If we have thousands of data,

06:32we can do that.

06:34And we can do 98% accuracy.

06:36As much accuracy, ma'am,

06:38our model will be more efficient.

06:40You have also said that

06:43if a forensic lab is used for this model,

06:47then a level will be very high.

06:50Those international labs

06:52can become a CD.

06:55So,

06:56until now,

06:57you have talked about

06:59when they want to use this model

07:01and they will work on this model.

07:03And they will work on this model.

07:05There is no such thing, ma'am.

07:07We will know that

07:08we will get copyright on Saturday.

07:10We will talk about something.

07:12And we will see.

07:13And in the international way,

07:14ma'am, this is our model.

07:15It is a work.

07:16Because it can happen.

07:17We cannot show that

07:18we have seen in literature

07:19that anyone has created

07:20such a face-created

07:21by using AI.

07:22And we have taken machine learning models

07:25that are quite unique models

07:26and prepared that

07:29Dhamni, our team has

07:32Rakesh.

07:33Akan Shah.

07:34Ayushi.

07:35And Akita.

07:37Nandini.

07:38They have worked very well

07:39by creating models.

07:40We have taken models.

07:41We are all our customized models.

07:43And now we have not done anything like that our model is...

07:48Because this is a government, we are all government.

07:51We do not want our model to be big.

07:53It is not a big thing, it is a big thing to use.

07:56We will not use it.

07:58It is not a Punjab University, it is a big model.

08:01So, the Punjab University is the owner.

08:03If someone wants to help,

08:06someone comes to help us,

08:08I am not ready for this.

08:10And it is a big model, it is a big model.

08:12And it is possible to give us a suggestion

08:14that you can improve it like this.

08:17So we can improve it too.

08:19But in the international way,

08:21we have not seen that anyone has prepared this model.

08:25And we will publish this paper too.

08:28So we will know how good it is published in our journal.

08:34So that is the biggest thing, ma'am.

08:37How is the value of our model?

08:42I'm my colleague from the University of Anthropology,

08:44Department of the University,

08:45Professor Keval Kishan.

08:47They have made a model for their team.

08:48They have created a model for their faces,

08:49only from their ears and fists.

08:59They can try to make this model more,

09:01and they will perform immediately.

09:02and they will do things quickly.

09:03and the department will quickly work on this model.

09:06With Camera Presentation, Nagma Singh report from ETV, Bhairat.

Category

Transcript

Recommended