Skripsi
PENGENALAN EMOSI MANUSIA MELALUI WAJAH DAN SUARA DENGAN MENGGUNAKAN CNN
These days, difficult jobs can be done by man-made objects called robots. One of them is a robot that resembles a human called a humanoid robot. To resemble a human, a humanoid robot should be able to recognize human emotions as much as possible, because emotions are common in humans in the form of happy, sad, angry, calm, and surprised emotions. So this study will develop a CNN model that can recognize emotions through a combination of faces and voices because previous studies only used one of the characteristics possessed by humans. To recognize emotions, this research uses CNN architecture in the form of VGG and AlexNet. By using the python programming language to train the CNN model, the results show that the model is able to recognize human emotions through the face or through the voice with the best accuracy value of 76.4% on the VGG facial model only and 75.9% on the combined model in recognizing faces with emotions. neutral. And the best accuracy on voice recognition can only read 100% angry emotions on the VGG voice model and 100% happy emotions on the combined model. However, this can be overcome by using voice input as well as the tensorflow model with an accuracy of up to 90% for the entire model. With this it can be concluded that the model can recognize human emotions with input in the form of faces and voices with the VGG architecture performing better than the AlexNet architecture.
Inventory Code | Barcode | Call Number | Location | Status |
---|---|---|---|---|
2207002537 | T75399 | T753992022 | Central Library (Referens) | Available |
No other version available