https://github.com/kimmandoo/faceemotionrecognition

FaceEmotionRecognition derived by lampadovnikita's archive. An Android application performing recognition of facial emotions on an image.
https://github.com/kimmandoo/faceemotionrecognition

android convolutional-neural-networks emotion-recognition face-detection face-emotion-recognition face-recognition kotlin mlkit-android mlkit-face-detection tensorflowlite

Last synced: 6 months ago
JSON representation

FaceEmotionRecognition derived by lampadovnikita's archive. An Android application performing recognition of facial emotions on an image.

• Host: GitHub
• URL: https://github.com/kimmandoo/faceemotionrecognition
• Owner: kimmandoo
• License: mit
• Created: 2024-10-28T13:04:58.000Z (about 1 year ago)
• Default Branch: main
• Last Pushed: 2024-10-28T13:31:56.000Z (about 1 year ago)
• Last Synced: 2025-04-03T20:33:48.079Z (9 months ago)
• Topics: android, convolutional-neural-networks, emotion-recognition, face-detection, face-emotion-recognition, face-recognition, kotlin, mlkit-android, mlkit-face-detection, tensorflowlite
• Language: Kotlin
• Homepage:
• Size: 4.13 MB
• Stars: 0
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

          
# EmotionRecognition

## This repository is derived by [lampadovnikita's archive](https://github.com/lampadovnikita/EmotionRecognition)

This repository represents an android application performing recognition of facial emotions on an image.

### The parts that I changed..

- My project is written by Jetpack Compose and Kotlin only. 

- No GPU use, only cpu tensorflow lite

- Not firebase ml vision, It was changed to MLkit's FaceDetection API

|  |  |

|-|-|

|![image](https://github.com/user-attachments/assets/d9dcd153-85a9-40d2-84db-229a5175c0be)| |

---

To train the CNN model there used hybrid dataset composed of the following datasets images:

- CK+ (all images except contempt images).

- JAFFE (all images).

- FER2013 (all images).

- RAF-DB (all images but only 205 happy class images).

  

The resulting hybrid dataset contains 46614 images and has the following data distribution:

All images was converted into the FER2013 images format - greyscale 48*48 pixels.

![image](https://github.com/user-attachments/assets/cfa64b8e-926e-49b9-bb6f-09b8bc502157)