About
Android application that solves the game Where's Wally? using Neural Networks.
Installation
The app can be installed using Android Studio.
First, clone the repository with:
git clone https://github.com/franksacco/where-is-wally.git
From Android Studio, select Import Project, then select the folder android
of the cloned repository.
Click Make Project to build the app and download all the required dependencies.
Click Run app to install the app on your device or emulator.
Screenshots
Performances
I installed the application on my phone and I ran some tests using 3 sample images to measure the time it took to locate Wally.
Below, I report the specifications for the phone and the sample images.
Device | Xiaomi Redmi 4 Pro |
---|---|
OS | Android 6.0.1 (Marshmallow), MIUI 10.2 |
Chipset | Qualcomm MSM8953 Snapdragon 625 (14 nm) |
CPU | Octa-core 2.0 GHz Cortex-A53 |
GPU | Adreno 506 |
RAM | 3 GB |
Image | Original size | Size with padding | Number of sub-images/tasks | |
---|---|---|---|---|
1 | 256 x 256 | 256 x 256 | 1 | View image |
2 | 2953 x 2088 | 3072 x 2304 | 108 | View image |
3 | 4130 x 2455 | 4352 x 2560 | 170 | View image |
Here I show you some interesting statistics about the tests performed.
Image | GPU | Parallel tasks | Total time | Tasks time | Average time per task | Max RAM used * |
---|---|---|---|---|---|---|
1 | ✖️ | 1 | 2.020 s | 1.790 s | 1.790 s | 345.2 MB |
1 | ✔️ | 1 | 2.735 s | 0.404 s | 0.404 s | 189.9 MB |
2 | ✖️ | 1 | 204.598 s | 160.106 s | 1.482 s | 424.2 MB |
2 | ✔️ | 1 | 91.729 s | 45.360 s | 0.420 s | 334 MB |
2 | ✖️ | 2 | -- | -- | -- | OOM ** |
2 | ✔️ | 2 | 80.873 s | 34.304 s | 0.318 s | 342.9 MB |
3 | ✖️ | 1 | 319.213 s | 250.332 s | 1.473 s | 495.0 MB |
3 | ✔️ | 1 | 142.445 s | 70.834 s | 0.417 s | 402.7 MB |
3 | ✖️ | 2 | -- | -- | -- | OOM ** |
3 | ✔️ | 2 | 126.331 s | 54.162 s | 0.319 s | 422.1 MB |
(*) Measured using Android Studio Profiler
(**) Out of memory