The answer from @Shaktiyavesh Nandan Pratap Singh has tackled an important point your application and how you prepare yourself for the work.

Nowadays, with so many packages/platforms and the need to produce results quickly, the basic apects related to image processing are not studied. People repeat things without discussing topics, taking into consideration deployment issues and without working as a reviewer.

The most important issue is to be willing to study and spend time developing something. The rest will follow.

Hi Mr @Sherief Hashima

These topics can be good research in the field of image processing:

1‌. Recognize traffic signs from images based on color and shape

2. Face recognition and its modes

3. Recognize the feeling of the face by extracting facial features

4‌. Online object detection and tracking of video images

5. Identify license plates and identify couples and individuals

6. Speed control of cars from video file

7. Diagnosis of diabetes

8. Target tracking in video images

9. Remove fog from foggy images

10. Recognize human movement

11. DCT detection of motion copy forgery

Good luck

Hi Mr. Sherief Hashima.

I firstly recommend@Majid Jangizehi. Since it is not very clear that in which area you are seeking for the AI application, so I would like to suggest the following measures based on your question:

1. Please select your sector of operations: Industrial, Social, Medical, Agriculture, Space, Aviation, Defence, Education, Engineering, Management etc.

2. Now , select the area of improvement/ process or interest wherein the AI application may be beneficial to improve the system and quality.

3. Next, you may seek the decision matrix to use various tools available with AI along with the Image Processing that you want to go with.

After following the above to narrow down your area of application, the followings may be suggested using AI with image processing: (For example: Aviation Sector)

a. In the management role like: processing of passes, passports, passengers identification, luggages/cargo management, tickets management system, etc.

b. In the technical roles like: flight data recordings, Maintenance, Fault Rectifications, Spares management, Rotables status, Prediction of failures etc.

c. In the Operation role like: Navigation system, Health Monitoring etc.

This is a type of frame work wherein you can actually explore the AI possibilities.

I would be delighted if you can share your area of interest wherein I may further add-on.

Thank You.

Hi @Shaktiyavesh Nandan Pratap Singh

You completed my answer well.

And more interestingly, I did all these steps for my dissertation on the aviation industry.

Thank you and good luck

Hello,

According to my view blockchain and AI both are emerging area. Research topics are how to integrate both the things in various applications.

We must take care when using training data because there is no guarantee that it will cover an unseen problem. Again the use of features to recognise patterns offers no guarantee that these features will be strongly present in an unseen problem. We must ignore results from theoretical problems like chess because they do not bear a relationship with reality.

A few directions:

- neural-symbolic hybrid approaches for semantic scene understanding in images/videos

- online training approaches for the already developed deep learning architectures for image proc tasks

- knowledge-based transfer learning for detection and recognition of unknown scenes in images/videos

- optimization of the deep learning methods for image classification, recognition, object detection etc

It will dominate over day-to-day security, medical, entertainment, banking and many other applications of various science. A brief light is here through : https://magnimindacademy.com/image-processing-and-its-future-implications/

Image processing based machine learning algorithms till now are based on low-level structural features. The next step in evolution would be directly recognizing high-level semantic features.

Developing the AI models for object recognition or object classification that can be implemented in small-type light-weight  hardware.

Due to the huge increase in multimedia, I think it is a good idea to think about reducing the processing time and efficiency. Selective information processing is one of the options, which means processing part of the scene. this can be implemented by simulating human vision system and human attention principles on image processing algorithms.

Best,

Mohammad

The answer from @Shaktiyavesh Nandan Pratap Singh has tackled an important point your application and how you prepare yourself for the work.

Nowadays, with so many packages/platforms and the need to produce results quickly, the basic apects related to image processing are not studied. People repeat things without discussing topics, taking into consideration deployment issues and without working as a reviewer.

The most important issue is to be willing to study and spend time developing something. The rest will follow.