Detectron 安装
基于 Caffe2.
Detectron 项目:
- Feature Pyramid Networks for Object Detection
- Mask R-CNN - Marr Prize at ICCV 2017
- Detecting and Recognizing Human-Object Interactions
- Focal Loss for Dense Object Detection - RetinaNet – Best Student Paper Award at ICCV 2017
- Non-local Neural Networks
- Learning to Segment Every Thing
- Data Distillation: Towards Omni-Supervised Learning.
- Faster R-CNN
- RPN
- Fast R-CNN
- R-FCN
基于的 Backbone 网络结构:
- ResNeXt{50,101,152}
- ResNet{50,101,152}
- Feature Pyramid Networks (with ResNet/ResNeXt)
- VGG16
Mask RCNN 输出.
From Detectron README.md.
1. Detectron 安装
Detectron 现还未给出 CPU 实现,需要基于 GPU 环境.
Requirements:
- NVIDIA GPU, Linux, Python2
- Caffe2, various standard Python packages, and the COCO API
- CUDA 8.0, cuDNN 6.0.21.
1.1 Caffe2 安装
参考 Caffe2 - (一)Source 安装及问题解决 和 Caffe2 官方 Install Instructions.
由于 Detectron 需要 Caffe2 包含 Detectron module,查看是否有该模块,没有的话更新 Caffe2 版本.
将 Caffe2 的 Python package 路径添加到 PYTHONPATH
环境变量,并确保 Caffe2 安装成功: