TRASH COLLECTION AND TRANSPORTATION

MONITORING SYSTEM

ABSTRACT: -

Supervising the process of garbage collection and transportation is a very crucial task with significance
implication to the reusing and the recycle of the garbage. Previous works mainly focus on the detection of
certain kind of garbage instead of overseeing the whole process. This paper proposed a supervision
approach based on the improved CNN modelto decrease the incidence of unexpected mixing of the
different kinds of garbage during the transportation and the collection which will cause the inferior
performance of waste classification. Firstly, to reduce the parameters and arithmetic operations of CNN
model and improving the model’s detection speed, we displaced the standard convolution in CNN model
by depth-wise separable convolution. Secondly, to solve the problem that CNN model has poor location
accuracy and performs poorly in muti-targets, triplet attention is introduced into the backbone, which
increases almost no parameters to automatically learn cross-dimensional interactions, enhance the
effective feature channel weights, and strengthen the feature extraction capability. Finally, we built a
dataset of waste classification supervision using the images provided by a city environmental protection
bureau on which we did massive experiments. The experimental result shows that compared with other
detection algorithms, the improved CNN model algorithm has better performance. The mAP is 99.91%,
which is 0.7% and 1.1% higher than the CNN modeland the average detection speed of the model is
14.6ms/it, which meets the requirements of regulatory real-time and environment complexity of the
supervision system.
EXSISTING SYSTEM PROPOSED SYSTEM
 A garbage image classification method  The CNN in Yolov3 is used to process the input
based on improved MobileNet v2 was image and extract features that are relevant for
proposed aiming at the problems of poor object detection. The CNN architecture in Yolov3
real-time performance and low consists of multiple convolutional layers and
classification accuracy of existing garbage pooling layers, which are used to extract
image classification models. A lightweight increasingly abstract features from the input
feature extraction network based on image.
MobileNet v2 was constructed.  The Yolov3 detection layer takes the output of the
 The parameter numbers of the model were CNN and applies a series of convolutional and
reduced by adjusting its width factor, fully connected layers to generate a set of feature
channel and spatial attention modules were maps that are used to make predictions about the
embedded in the model to enhance the objects in the input image.
network's ability to refine features, a multi-  The use of a CNN allows Yolov3 to learn spatial
scale feature fusion structure was designed hierarchies of features from the input image,
to enhance the adaptability of the network which enables it to detect objects with high
to scale, and transfer learning was used to accuracy and speed.
optimize the model parameters to further
improve the model accuracy. Experimental
results shows that the highestaccuracy.
EXISTING ALGORITHM PROPOSED ALGORITHM: -
 MobileNet v2  CNN model
 MobileNetV2 is a convolutional neural ALGORITHM DEFINITION: -
network architecture that seeks to perform  A CNN is a type of deep learning model that is
well on mobile devices. It is based on an specifically designed for image processing tasks
inverted residual structure where the such as object detection, image classification, and
residual connections are between the segmentation. It consists of multiple layers,
bottleneck layers. The intermediate includingconvolutional layers, activation
expansion layer uses lightweight depthwise functions, pooling layers, and fully connected
convolutions to filter features as a source of layers.
non-linearity.  The CNN model works by processing the input
 As a whole, the architecture of image through a series of convolutional layers,
MobileNetV2 contains the initial fully which apply a set of filters to the image to extract
convolution layer with 32 filters, followed features.
by 19 residual bottleneck layers.
DRAWBACKS: - ADVANTAGES: -
The problem of negative transfer. CNN model has the advantages of detection speed
Only works if the initial and target and accuracy.
problems are similar enough for the first This helps to reduce the computational complexity
round of training to be relevant. of the model.
It has a large number of backbone network
parameters.
SYSTEM ARCHITECTURE:

Fig:- proposed model

MINIMUMSYSTEM REQUIREMENTS
HARDWARE REQUIREMENTS

 PROCESSOR : Pentium i3 Processor

 RAM : 2GB DD RAM
 HARD DISK : 250 GB

SOFTWARE REQUIREMENTS

 BACK END : PYTHON

 OPERATING SYSTEM : WINDOWS 7
 IDE : Spyder3