EfficientNet-B0 Based Monocular Dense-Depth Map Estimation

Monocular depth estimation is a hot research topic in autonomous car driving. Deep convolution neural networks (DCNN) comprising encoder and decoder with transfer learning are exploited the proposed work for monocular map of two-dimensional images. Extracted CNN features from initial stages later upsampled using sequence Bilinear UpSampling layers to reconstruct map. The forms feature extraction part, image reconstruction part. EfficientNetB0, new architecture used pretrained weights as encoder. It revolutionary smaller model parameters yet achieving higher efficiencies than architectures state-of-the-art, networks. EfficientNet-B0 compared two other networks, DenseNet-121 ResNet50 models. Each these three models encoding stage followed by bilinear method decoder. an ill-posed problem thus considered regression problem. So metrics F1-score, Jaccard score Mean Actual Error (MAE) etc., between original reconstructed image. results convey that outperforms validation loss, F1-score ResNet-50