High-Level Project Summary
The plastic that ends up in the ocean accounts for 80% of the plastic produced annually, and it is about 500 million tons of plastic A reliable data set when generating the code to solve the problem, since it is based on the automatic learning method using deep learning and machine learning, we plan to create a drone that can fly in the air and dive into the depths of the ocean to detect marine debris. We chose this project because it represents a problem for many organisms besides the human element as the first to be affected, such as waste that cause to have some distortions marine animals and be the cause of their death.
Link to Project "Demo"
Link to Final Project
Detailed Project Description
Our model based on Deep Learning Network which composed of multiple automatic encoders. A deep network model uses sparse autoencoder adds a sparse constraint to the autoencoder, which is typically a sigmoid function. During learning, if a neuron is activated, the output value is approximately 1. If the output is approximately zero, then the neuron is suppressed. , network structure of the automatic encoder is shown in Figure 1. , basic principle of forming a sparse autoencoder after the automatic encoder is added to the sparse constraint as follows. It is assumed that the training sample set of the image classification is x(1) , x(2) , . . . , x(m), and x(m) is the image to be trained. Training is performed using a convolutional neural network algorithm with the output target y (i) set to the input value, y(i) x(i) . In Figure 1, the autoencoder network uses a three-layer network structure: input layer L1, hidden layer L2, and output layer L3. Its training goal is to make the output signal x approximate the input signal x, which is, the error value between the output signal and the input signal is the smallest. , number of hidden layer nodes in the self-encoder is less than the number of input nodes. If the number of hidden nodes is more than the number of input nodes, it can also be automatically coded. At this point, it only needs to add sparse constraints to the hidden layer nodes. In general, high-dimensional and sparse signal expression is considered to be an effective expression, and in the algorithm, it is generally not specified which nodes in the hidden layer expression are suppressed, that is, artificially specified sparsity, and the suppression node is the sigmoid unit output is 0. Specifying ρ sparsity parameter in the algorithm represents the average activation value of the hidden neurons, i.e., averaging over the training set. In node j in the activated layer l, its automatic encoding can be expressed as a(l) j.
asic Principle of Classification of Nonnegative Sparse Representation of Kernel Function. Premise that the nonnegative sparse classification achieves a higher classification correct rate is that the column vectors of D [v1, v2, . . . , vn] are not correlated. But in some visual tasks, sometimes there are more similar features between different classes in the dictionary. If the two types of problems are considered, the correlation of the column vectors of D1 and D2 is high, and the nonzero solutions of the convex optimization may be concentrated on the wrong category. A kernel function is a dimensional transformation function that projects a feature vector from a low-dimensional space into a high-dimensional space.
Our model, the output of the last layer of SAE is used as the input of the classifier proposed in this paper, which keeps the parameters of the layers that have been trained unchanged. Weights obtained by each layer individually training are used as the weight initialization values of the entire deep network. Fine tune the network parameters. Basic flow chart of the constructed SSAE model is shown in Figure 2.
Space Agency Data
I found in these sources high-resolution information, and I found it easy to understand, it is one of the strong and confirmed sources that can be relied upon to solve any problem, and I have gained a lot of very important information and information that I have never read, all of this I found in the sources of the Nasa website, and I will nominate it to my colleagues if they want to confirm the validity of some information
Hackathon Journey
Through my participation I can say that it was a unique experience to deal with colleagues from all over the world and then be on a team, it was a very useful experience for me and the spirit of teamwork that shows the love of science and the desire to benefit and protect humanity, in addition to taking advantage of the new information I read, as well as the tasks that have been carried out from me, all of which have benefited me a lot, and this tour is a wonderful kind of work The group that gives many gains, such as: experience, understanding problems, how to think, choosing the right solution, working to implement it, insight into the planet and the risks and problems it faces and trying to reach the right solution as soon as possible.
References
https://paperswithcode.com/paper/the-marine-debris-dataset-for-forward-looking
https://paperswithcode.com/paper/deepplastic-a-novel-approach-to-detecting
https://conservancy.umn.edu/handle/11299/214865
https://github.com/antiplasti/Plastic-Detection-Model
https://www.epa.gov/trash-free-waters/toxicological-threats-plastic
https://github.com/facebookresearch/detectron2
https://colab.research.google.com/drive/16jcaJoc6bCFAQ96jDe2HwtXj7BMD_-m5#scrollTo=PIbAM2pv-urF
We use google colab to write the code and run it .
https://git-scm.com/
https://colab.research.google.com/drive/16jcaJoc6bCFAQ96jDe2HwtXj7BMD_-m5#scrollTo=PIbAM2pv-urF
We use google colab to write the code and run it
Tags
#hardware , #marine debris
Global Judging
This project has been submitted for consideration during the Judging process.