Age Range Estimation using MTCNN and VGG-Face Model
This work addresses age estimation for applications like security or marketing, but it is incremental as it builds on established techniques like MTCNN and VGG-Face.
The authors tackled age range estimation from face images by using MTCNN for face extraction and a VGG-Face model with transfer learning, achieving performance that outperformed existing state-of-the-art methods on the Adience benchmark.
The Convolutional Neural Network has amazed us with its usage on several applications. Age range estimation using CNN is emerging due to its application in myriad of areas which makes it a state-of-the-art area for research and improve the estimation accuracy. A deep CNN model is used for identification of people's age range in our proposed work. At first, we extracted only face images from image dataset using MTCNN to remove unnecessary features other than face from the image. Secondly, we used random crop technique for data augmentation to improve the model performance. We have used the concept of transfer learning in our research. A pretrained face recognition model i.e VGG-Face is used to build our model for identification of age range whose performance is evaluated on Adience Benchmark for confirming the efficacy of our work. The performance in test set outperformed existing state-of-the-art by substantial margins.