D
D VarunTeam Update:
We initially got the accuracy of 89.9% for 2000+ image datasets for our Convolutional Neural Network (CNN). Later we attained an accuracy of 96.5% for 10000+ image datasets for the same CNN.
The image Dataset included all kinds categories of images like satellite images, long chain of plastic, plastic floating under the surface few meters below, surface floating plastic, plastic near the sea shores and many more. Therefore we have trained with a enormous varieties of images to accomplish a highly accurate network for classification, detection, segregation, decision and finally elimination of plastic from the ocean if it is detected as a marine plastic debris.
Hence as above mentioned is the flow of process if the network is fed into any hardware model or a robot.
Below are the graphs obtained from testing and training processes:
D
D Varun