SeeFood App

Getting Started

These instructions will get you a copy of the project up and running on your mobile for development and testing purposes.

Prerequisites

What things you need to install the software and how to install them

An Android mobile device with a minimum SDK of 18.

Internet connectivity.

Installing

• Running on Android Device

  1. Download and unzip project folder from GitHub
  2. Connect Android device to computer via USB cable and enable file sharing
  3. Copy the release APK from the release folder within the root directory of the project into a desired folder on the Android Device
  4. Using the Android Device, click the APK file and give the app permissions to install

Deployment

Once installed onto the Android Device, the SeeFood app can be deplyed through the device's app menu by clicking on its icon.

Note

If issues are experienced during installation, make sure that the Android device is able to install apps from Unknown Sources. This setting can be changed at Settings->Security and checking the option under Device Administration.

Using the App

The SeeFood app allows users to submit images to a server hosted through AWS (Amazon Web Services). The server uses a pretrained neural network to assess whether the submited images contain food. A result screen is displayed after image submission, which shows the user the confidence of the neural network's decision and its answer.

Design

This app was designed using Adapter Design Patterns, several additional classes, and an additional repository to store and modify the AWS server. The app was developed in modules assigned to each member of the team. Ibrahim was responsible for the UI and the ability to submit images. Brittany was responsible for the the gallery and uploading/downloading its contents. Rafael was responsible for maintaining and managing the server.

Classes created to manage the UI:

• Feedback_Window:
  • This class is responsible for the displaying the selected or captured image along with the confidence rate produced by the AI. Thus, Feedback_Window class is executing multiple functionalities that were specified for the application.These functionalities include connecting to the AI via EC2, uploading images(s) to the server, and graphically displaying the dfidence result transmitted by the server. In addition, the Feedback_Window class will prompt the user about any possible issues that could arise before or during the AI processing conducted on the server.

Classes created to manage the Gallery:

• DepthPageTransformer:

  • This class was a custom transformer to be used with the viewPager in the gallery. This class adjusts the image fragments it is applied to in order to inflate the image to a slightly larger size when it is opened by the user for horizontal scrolling through the gallery.

• DetailActivity:

  • This class is an Activity that creates the viewPager in the gallery such that the user is able to select an image, it will be inflated via the DepthPageTransformer, and the user is able to swipe horizontally(left or right) through all of the images in the gallery. This class also offers the ability of the user to select an image while in the viewPager for selection in order to view its stats as determined by the AI.

• DownloadFileAsync:

  • This class extends AsyncTask in order to download the gallery from the server as a zip file. It opens the web connection for the server and initializes and completes the download of the folder and its contents using streams. Note, that this class completes the download of the zip folder in th background of the application. Thus, when this class is reached during execution of the gallery functionality, the app waits until the gallery has been fully downloaded before commencing actions such that the entire folder is successfully downloaded first.

• Gallery:

  • This class is an Activity that executes the download of the zip file, unzips the folder and extracts the images onto the device, populates the gallery with said images, and executes the listener that opens the viewPager for swiping through the gallery. This is how the images are displayed as a grid to the user. Additionally, this class also deletes the zip folder and the directories containing the pictures regularly, such that the gallery is always up-to-date for the user.

• GalleryAdapter:

  • This class is a custom adapter for the RecyclerView used to display all of the images within the gallery as GalleryItem objects. The class displays the images in the modified RecyclerView in a grid using Glide as the image loaded library.

• GalleryItem:

  • This class defines the GalleryItem object by assigning a name and url to each of the images stored within th gallery. This is useful for calling the FeedBack_Window object, which requires the Uri of the image.

• ImageClickListener:

  • This class is a custom listener that opens the viewPager should the user select any image in the GalleryAdapter. The image selected will be inflated to a slightly larger size by the DepthPageTransformer, and the images can be swiped through via the DetailActivity.

Built With

• Gradle - The building system used.
• Glide - Gallery library used.
• AWS - Web services.

Versioning

We use Github for versioning. For the versions available, see the tags on this repository.

Authors

• Brittany Woods - Github account
• Ibrahim Almohaimeed - Github account
• Rafael Nunez - Github account

License

This project is licensed under the Apache 2.0 License - see the LICENSE file for details.

Acknowledgments

• A tutorial created by Suleiman was used to create the Gallery and implement the viewPager on image selection.
  • https://blog.iamsuleiman.com/image-gallery-app-android-studio-1-4-glide/
  • https://github.com/Suleiman19/Gallery