Inspired by the 'SqueezeNet' proposed by Forrest Iandola et al. (2016) architecture, created a smaller model for CIFAR-10 data set using similar components (fire module, etc).

Object Detection in Twilight

Machine learning project for Computer science (University of Catania)

Easy to use, popular computer vision layers' implementations with customizable parameters

Automatic Diagnosis of COVID-19 using CT Scan

Steering Angle prediction

Deep learning models to predict the emotion exhibited by the person from videos

Model compression methods applied to the monocular depth estimation network by Godard et al.

Malayalam Sign Language Recognision

Taken from https://answers.opencv.org/answers/191837/revisions/

This repo shows AI Implementation and how to use ONNXJS with Next.js.

the implementation of some deep learning algorithms

Python program allowing to control mouse by performing hand gestures captured by a webcam.

Behavior cloning of driving a car on a track

Facebook AI Performance Evaluation Platform

How AI enhances parking management. In a team of four, we leveraged deep CNNs to create a system that detects (YOLO) and classifies (ensemble of SqueezeNet and ResNet) parking spots as free or occupied from images obtained by common surveillance cameras. It is able to employ pre-installed cameras, optimizing space utilization and reducing costs.

Cpp image classification with Caffe and OpenCV

Image Scene Classification of Multiclass using transfer learning

Implementation of SqueezeNet: AlexNet-level accuracy with 50x fewer parameters and <0.5MB model size by Forrest N. Iandola, Song Han, Matthew W. Moskewicz, Khalid Ashraf, William J. Dally, Kurt Keutzer.