/OpenCL-VGG16

OpenCL based VGG16 implementation

Primary LanguageC++MIT LicenseMIT

OpenCL_VGG16

Using the OpenCL framework, we built a VGG16 CNN(Convolutional Neural Network) for object recognition to run on heterogeneous environments such as CPUs, GPUs, and FPGAs. OpenCL provides hardware acceleration similar to CUDA, but it can be deployed on non-Nvidia GPUs, making it widely used for edge and fog computing.

To-Do

  1. Test Output for each layers. Navneet
  2. Create class for Dense Praphat
  3. Create class for Max Pool. Avneet
  4. Creating complete network.
  5. Python script to convert Dense weights to txt. Prabhat Image to txt. Avneet