Kamran B. Profile

UDFORSK

C Programming

Computer Security

Engineering

JavaScript

Python

0,0

(0 anmeldelser)

Du følger nu .

Fejl under følg af bruger.

Denne bruger har ikke tilladt andre bruger at følge dem.

Du følger allerede denne bruger.

Din medlemsskabsplan tillader kun at følge 0. Opgrader her.

Holdt op med at følge bruger.

Kunne ikke stoppe med at følge bruger.

Du har nu anbefalet

Kunne ikke anbefale bruger.

Noget gik galt. Forny venligst siden og prøv igen.

Email verificeret.

$30 USD / time

istanbul,

turkey

$30 USD / time

Det er i øjeblikket 6:07 AM her

Tilmeldt februar 26, 2017

0 Anbefalinger

Kamran B.

@BalayevKamran

0,0

(0 anmeldelser)

0,0

$30 USD / time

istanbul,

turkey

$30 USD / time

N/A

Jobs færdiggjort

N/A

Indenfor budgettet

N/A

Til tiden

N/A

Genansættelsesrate

AI & Software Developer

I have graduated from Computer Science department from YTU. When i was high school, I started web programming and carried out various projects. While doing projects on the web programming, I also started to learn the field of artificial intelligence when i was started university. I have developed various projects with AI, like classification, regression, creating a special data extractor from image dataset, a synthetic data generator from image dataset, program which creates captions from images and also finding the best caption for an image and etc. I have also published an article of synthetic data generator project. When I was student i had the opportunity to gain coorporate experience in the computer vision department of defense industry and other companies. When I was the third year of my university, I established and incorporated my own AI based startup. I am actively working in my own company as a advisor. My main purpose in creating this profile is to spend some of my free time by following current technologies and participating in development of different projects because programming is not only work for me also it is my hobby :)

Se alle C Programmers i Turkey

Kontakt Kamran B. angående dit job

Log ind for at diskutere detaljer over chat.

Portefølje indlæg

A large amount of data is the key requirement in order to train a neural network efficiently. Using a small size training set in network training causes low accuracy for model performance over the testing set and also hard to implement the model in practice. Similar to many other problems, sperm morphology datasets are also limited for training the neural network-based deep networks in order to
provide an automatic evaluation of sperm morphometry. Data augmentation mitigates this problem by utilizing actual data more effectively.

The standard data augmentation techniques focus on only spatial changes over the images and can only produce a restricted number of useful informative and disjunctive data. Therefore, in order to create more distinctive and diverse data than the regular spatial domain-based augmentation techniques, a deep learning-based data augmentation technique which is known as the generative model, is trained in this study for the sperm morphology datasets.

The Implementation of DCGAN in the Data Augmentation

In this project, we present generative and contrastive models which are used to create captions and classify photos according to user input. Models have been trained on several datasets such as Flickr and COCO. We also trained our models on custom Tweet and News datasets which we scraped over 2 million consisting of image-caption pairs. We utilized the CLIP network of OpenAI to classify images given the text input. Our models for generating Turkish captions showed the BLEU scores as high as 45 whereas the classification results showed 97% prediction accuracy. At the end of the study, we show all the BLEU score results of the Flickr datasets, COCO dataset, Tweet and News datasets and compare the results between each of the trained model.

Turkish data labelling

Anmeldelser

Ændringer gemt

Ingen anbefalinger at se her!

Uddannelse

Computer Science

Yildiz Teknik Üniversitesi,

Turkey

2017 - 2022

(5 år)

Publikationer

The Implementation of DCGAN in the Data Augmentation for the Sperm Morphology Datasets

https://www.researchgate.net/

A large amount of data is the key requirement in order to train a neural network efficiently. Using a small size training set in network training causes low accuracy for model performance over the testing set and also hard to implement the model in practice. In order to create more diverse data than the regular spatial domain-based augmentation techniques, a DL-based data augmentation technique which is known as the generative model, is trained in this study for the sperm morphology datasets.