Object detection is a very important basic research direction in the field of computer vision and a basic method for other advanced tasks in the field of computer vision. It has been widely used in practical applications such as object tracking, video behavior recognition and underwater robotics vision. The Cascade-RCNN and Deformable Convolution Network are both classical and excellent object detection algorithms. In this report, we evaluate our Cascade-DCN based method on underwater optical image and acoustics image datasets with different engineering tricks and augumentation.