Computer Vision Projects

Technologies

Platforms

Projects

ID	Project	Description
0001	CNN From Scratch Tenserflow	Building a Simple CNN Model in Tensorflow (First time experimenting CNNs in Tensorflow).
0002	CNN Fashion Image	Building a CNN Model for the Fashion-MNIST Dataset.
0003	CNN Hand Gesture	Building a Hand Gesture Recognition for American Sign Language.
0004	Facial Expression Recognition	Facial Expression Recognition on the FER Dataset.
0005	Image Colorization	Image Colorization from black and white photos to colored photos.
0006	Audio Classification	Audio Classfication with the Gtzan Dataset.
0007	Facial Key Points Detection	Building a system to identify key facial landmarks.
0008	OCR Recognition	Optical Character Recognition using Pytorch
0009	Scene Classification	System built where scenes from photographs are categorically classified.
0010	Traffic Sign Recoginition	Traffic Sign Recognition.
0011	Real Time Shape Contour Detection	Detect contours of shapes from a live video feed.
0012	Real Time Color Tracker	Detect and Track color from the HSV scale from a live video feed.
0013	Object Detection with HaarCascade	Detecting Faces with the HaarCascade Filter.
0014	Hand Tracking Volume Adjuster	Program that allows to adjust the speaker volume of your Laptop or Desktop using your hand movement.
0015	AI Attendence Marker using Face Recognition	Attendance marking system with a live feed.
0016	AI Curl Trainer	Curl Trainer built using a pose detector.
0017	Angle Finder	Angle Finder from a live feed.
0018	Annotating Images with Object Detection API	Annotating Images and Videos using multiple Object Detection models like SSDs, RCNNs, RFCNNs.
0019	Bar and QR Code Detection	Detecting Bar Codes and QR Codes from a live video feed and image.
0020	Document Scanner	Document Scanner from a live feed.
0021	Flickr8 Image Caption Generation	Image Captioner built using Flickr 8k Dataset.
0022	Space Radio Signals Classification	Space Radio Classification on SETI Data.
0023	Cell Instance Segmentation	Instance Segmentation on the Sartorius Cell Instance Segmentation Dataset.
0024	Google Street View House Number Recognition	House Number Recogition on the The Street View House Numbers (SVHN) Dataset.
0025	Object Dimension Measurer	Measure Object Dimensions from a live feed.
0026	Optical Mark Recognizer	Optical Mark Recognizer(OMR) from a live feed.
0027	Text Detector Tesseract	Text Detector built using Google's Tesseract Engine.
0028	Toxic Comments Classification	Toxic Comment Classification with 1D Convolutions.
0029	Virtual Keyboard	Virtual Keyboard that can be used with a live stream.
0030	Number Plate Detection with EfficientDet	Number Plate Recognition on the CCPD (Chinese City Parking Dataset, ECCV) using the EfficientDet Model.
0031	Wheat Detection with EfficientDet	Wheat Grain Detection using the EfficientDet Model.
0032	COTS Detection with EfficientDet	Crown-of-Thorns Starfish (COTS) Detection in the Great Barrier Reef using EfficientDet.
0033	SIIM-FISABIO-RSNA COVID-19 Detection With EfficientDet	Identifying and localizing COVID-19 abnormalities on chest radiographs using the EfficientDet Model.
0034	VinBigData Chest X-ray Abnormalities Detection With EfficientDet	Automatically localize and classify thoracic abnormalities from chest radiographs using the EfficientDet Model.
0035	NFL 1st and Future-Impact Detection With EfficientDet	Detect helmet impacts in videos of NFL plays using the EfficientDet Model.
0036	Oxford-IIIT Pet Dataset Classification With EfficientDet	Classification of pets into 37 categories using the EfficientDet Model.
0037	Aquarium Dataset Classification with EfficientDet	Classification of water species using the EfficientDet Model trained on the "Aquarium Dataset".
0038	Crowd Detection With EfficientDet	Crowd Counting/Detection using the EfficientDet Model.
0039	Herb Detection With EfficientDet	Herb Plants Detection using the EfficientDet Model.
0040	Swimming Pool Detection With EfficientDet	Swimming Pool Detection on Aerial Images using the EfficientDet Model.
0041	Face Mask Detection With EfficientDet	Face Mask localization and Detection using the EfficientDet Model.
0042	Tracheostomy Tubes Detection With EfficientDet	Tracheostomy Tubes localization and Detection on X-rays using the EfficientDet Model.
0043	Barcode Detection With EfficientDet	Barcode Detection using the EfficientDet Model.
0044	Gotukola Detection With EfficientDet	Detecting and Localizing Gotukola leaves using the EfficientDet Model.
0045	Helminths and Schistosoma Mansoni Detection With EfficientDet	Detecting and Localizing different species, such as roundworms, whipworms, and hookworms, which causes Soil-transmitted Helminths and Schistosoma Mansoni infections using the EfficientDet Model.
0046	Fire Detection With EfficientDet	Fire and Smoke Detection using the EfficientDet Model.
0047	Semantic Segmentation with AWS Sagemaker	Semantic Segmentation of Pet Data with AWS Sagemaker.

Research Papers

Paper ID	Title	Description
0001	EfficientDet :Scalable and Efficient Object Detection	EfficientDet is a type of object detection model, which utilizes several optimization and backbone tweaks, such as the use of a BiFPN, and a compound scaling method that uniformly scales the resolution,depth and width for all backbones, feature networks and box/class prediction networks at the same time.
0002	YOLACT: Real-time Instance Segmentation	YOLACT is a real-time instance segmentation system that is capable of detecting multiple objects and segmenting them in an image or a video stream. It achieves this by using a one-stage detector, which is a type of deep neural network that can detect objects in a single pass through the image, without requiring a separate region proposal step. The system also employs a fully convolutional mask head, which is another neural network that generates instance masks directly from the detector's output. By doing so, the system can perform instance segmentation in real-time while maintaining high accuracy.
0003	YOLACT++: Better Real-time Instance Segmentation	YOLACT++ is an improved version of the YOLACT (You Only Look At CoefficienTs) model, which is designed for real-time instance segmentation tasks. YOLACT++ uses a modified version of the ResNet-101 backbone architecture combined with a Feature Pyramid Network (FPN) and a Panoptic Feature Pyramid Network (PFPN) to extract features from different scales. It then uses a lightweight prototype projection head that generates mask coefficients for each instance, which are used to construct instance masks in real-time.