
Deep Neural Projects
(1st part of the course: Images)
There are 13 projects in total for the first part of our course. All of them are somehow related to image processing, since the first part of the course is devoted mainly to computer vision.
There is no opportunity to submit your project this semester.
Unless otherwise specified, the project implies work in Jupyter Notebooks, and, if desired, the implementation of the functionality in the form of a web demo, mobile application or bot.
1. Image Search Engine

In this project, you will implement your own mini image search engine.
There is some database of images (it will be collected by you and your mentor for the project), a new image is input and you need to display the most similar images from the database
(say, top 10 similar ones).
2. Face Recognition

In this project, you need to implement a face recognition system. There is some database of images with people's faces for 5-10 Human. It is necessary to say from the input image whether this person is one of those 5-10, and, if so, what is the probability of this.
3.Face Keypoint Detector

An image of a person's face is received at the input, and it is necessary to detect points of his face
(it will be predetermined which ones, there will be a fixed number of them).
4. Image Captioning

In this project, you need to implement a neural network that prints a text description of the input image. Examples of work can be seen in the images above.
5. Segmentation in Medicine

In this project, you will need to write a neural network that qualitatively segments medical images, highlighting regions of the desired type on them.
6. 3D Pose Estimation

Here you have to write a neural network that predicts from the image
human pose in 3D. There will be a fixed set of body points,
it is necessary to accurately predict all of them and connect them with lines, obtaining a three-dimensional skeleton.
7. Text Extraction

In this project, we need to work on extracting text from images.
The input is a picture, you need to recognize regions with text in the picture
and display text from them on the screen.
8. Age Recognition

Predict a person's age by looking at a person's face.
9. Image Question Answering

The input is an image and a question. For example: "How many children are in this picture?".
Need to know exactly Answer this question with an image.
10. Gesture Action Recognition

In this project, you need to predict the position of the hands from the image, namely: the position of all phalanges of the fingers and the lower points of the hands. This should be done as accurately and quickly as possible, because potentially this technology can be used to control gestures. in real time.
11. Classification Bot


The project is to write a bot (in Telegram or VK), which, according to the image sent to it, will issue the class of the depicted object.
12 Style Transfer Bot


The project is to write a bot (in Telegram or VK), which, according to the sent him style images and photos will return a picture with transferred to the photo style.
13 Style Transfer Filters

In this project, the emphasis is on the implementation of the fastest and most high-quality transfer of style to photographs of people, the creation of so-called "filters".