Ge force titan x

All experiments are accomplished on a high-performance workstation in which two Intel Xeon CPUs are configured. Each CPU contains eight physical cores with hyper-threads technique, and the core clock rate is 3.1GHz. The host memory is 128GB. The GPUs in our experiments are the NVIDIA Ge Force Titan X (Maxwell architecture) and Tesla K10 (Kepler Architecture). The Ge Force Titan X contains one GM200 GPU with 3072 cores. The core clock is 1.0GHz. It is accommodated with 12GB device memory. The Tesla K10 includes dual GK104 GPUs, and each GPU contains 1536 cores, and the core clock rate is 745MHz. The device memory for every GPU in Tesla K10 is 4GB.

We train our model for 5 epochs with a batch size of 64. The size of input patches is 64 × 64, and the number of training patches is 120000. For each module, the numbers of feature channels are 16, 32, 64 for corresponding stages. All networks are trained using Adam optimizer with a learning rate of 0.0001. Dropout rates of all dropout layers are set to 0.2. For our proposed network, we compile the model and assign a weight of 1 for all dice coefficient loss, and 0.01 for overlap-loss. The model achieved a validation IoU of 0.793 averaged across the 6 outputs and a validation IoU of 0.97 on the combined reconstruction. All experiments were conducted on two NVIDIA GeForce Titan X (Pascal) GPUs.