A new clustering algorithm called HCADBSCAN was purposed which can significantly reduce the computational complexity of the DBSCAN algorithm while maintaining its accuracy. HCA-DBSCAN uses grid to create various clusters and merges them if they are similar to each other by a threshold value given by user.
This paper states the use to identify the faces in an image having differently orientated faces. Scores are generated by providing heat maps (caffe library) for each image which enables to detect the face. Once the heat maps are generate for each image, we create a model using the training set where each face has its own model. Hence when we use the test data, we scan for new faces and tag the label using the training model. If the face is not recognized, the user is asked to enter the label manually and it is then updated in the database.
We propose a system which is attached to government buses or even on private vehicles on the belly which have the following sensors: Accelerometer, GPS Module and Ultrasonic sensor. This helps us in feteching real time information about potholes which is send to the cloud at intermediate checkpoints. At the server end, data can be analysed which can be used to improve the quality of the road.