AttM-CNN: Attention and metric learning based CNN for pornography, age and Child Sexual Abuse (CSA) Detection in images

•A deep learning framework for facial age, pornography and child sex abuse detection.•A new CNN with visual attention mechanism and metric learning named as AttM-CNN.•Introduces a new dataset with human labeled 80k images with facial age for juveniles.•Presents another new dataset with 2 million adult pornographic images.•Comparative analysis on ImageNet, age & pornography benchmark datasets and real CSA images.