It’s Not Polite To Point: Describing People With Uncertain Attributes

It’s Not Polite To Point: Describing People With Uncertain Attributes

Visual and pictorial attributes are most attractive and powerful features in computer visionary software’s for the detection of some object or recognition of the scene. Besides the visual attributes of the computers, now, they also provides another attributes that is the verbal communication that has been not yet been examined in any of the provided researches.

It is important in any system to deal with the uncertainty issues. In this research the issue of communication has been addressed by analyzing and fixing the problem and issue of the tasks that has by generated by the computers by composing some programmed description about some individual in a photo, that marks distinguished him/her from the other people in the photo.

In this research a principled and well organized method has been chosen to reflect that what the chances that attribute giving short descriptions will refer and maximize the chances that another party will identify the person whose description has been given. In this research, the comparison has been made between algorithm and baselines of computer and this shows the strength of this method.

The center focus and area of concern of this research is to make the contribution of the computer as effective but not to make the contributions of computer more efficient and informative than they are required.

In the context of this research, where the computer will refer and describes a certain person, it has been interpreted as follows; principally the description that has been provided by the computer is targeted on the first person. Secondly, the description should be as concise and as brief as possible. The task of describing people is easy for the individual of the society but it is not for the computer.

In this kind of system, computer must deal and handle the uncertainty. The attribute classifiers of this computer system are loud and noisy. And for an effective model this problem of uncertainty should be dealt.

On the other hand, each person and individual has many attributes which will describe him and describe him uniquely.

For instance: a brute force system and method. In this method, first the descriptions have been tried with one attribute, and then two attributes has been tried and the method has been developed like this.  Even though this method will find the shortest explanation, the level of computational difficulty and trouble is exponential in the number of available attributes. This method has addressed and represents a significant part of the problem of uncertainty, and this research address the problems that can be faced in case of generating the short description about the images and the individuals present in it. Having the expression and its generation is one of the most initial building stone and block for any system that has natural language. This research is quite useful and has practical applications.

The application of this method is that it can identify the suspicious people through the surveillance and security camera, through the action recognition method also. This method will help the security guards to receive and use the verbal explanation that has been received by the computer to investigate and caught the suspect.

Both properties of the portrayal are tremendously crucial. First, the explanation and portrayal needs to mention only to the suspect to avert and stop investigating the erroneous person. Secondly, the description must not be too extensive as to confuse the guard or waste the time and resources by investigating the wrong person.

One more application is the navigation methods and system. By using a front-facing or selfie camera in a vehicle and a GPS system, moreover a system can also be developed which will provide more spontaneous driving directions.

For instance, instead of directing that “Turn right in 200 feet,” it will be more appropriate and useful to say: “Turn right at the yellow building with the red awning”.  Even though in this research we have used our procedure for describing people, it is not limited and narrowed to this precise domain only.  By engaging and retaining the object exposing algorithms, in accumulation to other attribute classifiers, it is also an advantage.

Our main contributions in this research are:

• This research presents the first attempt in producing and referring expressions of the objects in images and the attempt to describe them.

• .In accumulation to that, the research also presents a novel, modern and computationally advance and efficient method for estimating the likelihood and prospect that a given description provided by the computer will result in an accurate guess and will be helpful in processing the information.
• Finally, this research has also developed and establishes a new algorithm and procedure for attribute selection which takes into contemplation about the vagueness and improbability of the classifiers.
• Like every method and every procedure anything cannot guarantee d and guaranteed 100 percent that the description composed by the system will describe only the suspected and targeted person.

For future framework, this method and work can be extended and further researched to consider back and forth discussions between the computer systems and the humans. That is, if the referring expression and description provided isn’t clear, the guesser and the user will be allowed to ask certain specific question from the system to have more clarity.

This could involve responding to the user’s clarifying question, or providing response and reaction to a user who guessed incorrectly. Finally and lastly it has been believed that our framework is an important component for any image description algorithm, though challenges remain dealing with integrate more general image.