A GPU Implementation of GoogLeNet

Download

• Code: Patch to Caffe to train using a single GPU
• Trained model using ImageNet
• Trained model using Places

Acknowledgement

We thank Shi Qiu and Yuanjun Xiong for discussion on GoogLeNet implementation. We thank Antonio Torralba, Aude Oliva, Bolei Zhou and Agata Lapedriza for help on the experiments of Places database.