ajayansaroj17,Ajayan Saroj,sarojajayan1,github

Hi there 👋 I am Ajayan!

I have been working in Deep Learning, and Computer Vision-based models.Expertise in model development to deployment lifecycle.

I have worked on imaging and computer vision projects that involve classification, object detection, segmentation and tracking.During my graduation I have publised a paper on Smart Switch Board Compatible With Google Assistance Along With Face Recognition And Security System

Moreover, I'm also a Teaching assistant at IIT- Roorkee (Machine Learning). I am working on Generative Adverserial Networks (GANs) under guidance of Dr.GN Pillai (HOD - Electrical Engineering, Joint Faculty Mehta Family School).

My Interests are:

"Machine Learning, Deep Learning, Computer Vision, Statistical analysis"

Here is a bit about my interests and how to get in touch:

🔭 I’m currently working on Generative Adverserial Network.
💬 Ask me about new machine learning techniques and let's collaborate on making them even better!
📫 How to reach me: [email protected]
😄 Pronouns: he/him/his

Ajayan Saroj's Projects

ai_for_investment

To study and build various AI model for investment

ajayansaroj17

Config files for my GitHub profile.

american-sign-language-detection-using-cnn

ASL provides the deaf community a way to interact within the community itself as well as with the outside world. However, not everyone knows about signs and gestures used in sign language. With the advent of Artificial Neural Networks and Deep Learning, it is now possible to build a system that can recognize objects or even objects of various categories. The Kaggle ASL alphabet dataset was used to develop the CNN model, which has an 82% classification accuracy. The neural network is built using the Tensorflow framework, with CNN layers added initially and ANN layers added after.

comparing-vit-with-resnet-gradcam-with-attention

A basic CNN (ResNet18) and a vision transformer were trained using a dataset. Evaluated each's accuracy and showed which was better. Grad-CAM was also used on CNN to extract the attention maps of the ViT.

kmedoids

[Unmaintained] The Python implementation of k-medoids.

musical-notes-procreation-using-deep-learning

Basically, it is a time series problem wherein first we have to encode traditional musical notations (MIDI or abc notation) into numerical one and feed it into LSTM and subsequently predict the next notes which is in the numerical form. then after we have to decode the numerical values into respective notations either MIDI or abc notation. we can easily convert MIDI or abc notations into mp3 format and check the quality of the generated sound.

road-lane-detection

Created a road lane identification model that can distinguish between the several road lanes included in an image using FCN architecture and a pre-trained VGG-16 model without top layers for semantic segmentation.

Recommend Projects