这个仓库旨在实现常用的目标检测算法，主要参考如下：
- mmdetection，参考CUHK检测框架的思路。
- ssd.pytorch
- Object-Detection
- 2015-10-09-object-detection.md handong收集的相关目标检测论文目录；
- awesome-object-detection，awesome系列，参考合并了handong相关目录；
- 目标检测 Object Detection 博客整理收集的相关资料；
- Review of Deep Learning Algorithms for Object Detection 相关目标检测DL算法综述；
- deep_learning_object_detection，其中增加了可视化paper表格，非常直观；
- Deep Learning for Generic Object Detection: A Survey，深度学习通用目标检测调研；
- ...
- DeNet: Scalable Real-time Object Detection with Directed Sparse Sampling 相关代码denet。
- Soft Proposal Networks for Weakly Supervised Object Localization 相关代码SPN.pytorch
- ICCV 2015 Tutorial on Tools for Efficient Object Detection ICCV 2015中举办的关于目标检测的教程，可以参考。
- Deep Learning for Objects and Scenes CVPR 2017关于目标检测的教程。
- RSA-for-object-detection-cpp-version RSA-for-object-detection 相关论文Recurrent Scale Approximation for Object Detection in CNN
- DetNet: A Backbone network for Object Detection
- 小目标检测，参考如下
- 遮挡目标检测，参考如下
- 视频目标检测，参考video_obj
- domain目标检测，参考如下
- 非极大值处理，参考如下
- 弱监督目标检测，参考如下
- 困难样本采样策略，参考如下
- 文本检测，参考如下
- 类别不平衡目标检测，参考如下
- 小数据集目标检测，参考如下
- A unified multi-scale deep convolutional neural network for fast object detection
- How Far are We from Solving Pedestrian Detection? 行人检测
- Taking a Deeper Look at Pedestrians 行人检测
- Integralchannel features 行人检测
- Fast Feature Pyramids for Object Detection 行人检测
- What Can Help Pedestrian Detection? 行人检测
- Citypersons: A di- verse dataset for pedestrian detection 行人检测数据集
- DenseBox: Unifying Landmark Localization with End to End Object Detection，不使用anchor的检测方法
- UnitBox: An Advanced Object Detection Network
- Discriminative models for multi-class object layout
- Learning Transferable Architectures for Scalable Image Recognition，自学网络结构；
- Single-Shot Refinement Neural Network for Object Detection，S3FD和RefineDet论文都是同一个作者；
- HyperNet: Towards Accurate Region Proposal Generation and Joint Object Detection
- Hypercolumns for Object Segmentation and Fine-grained Localization
- Feature-Fused SSD: Fast Detection for Small Objects
- Perceptual Generative Adversarial Networks for Small Object Detection，GAN小目标检测，暂且不看；
- Detecting and counting tiny faces
- Seeing Small Faces from Robust Anchor's Perspective
- Face-MagNet: Magnifying Feature Maps to Detect Small Faces
- Small-scale Pedestrian Detection Based on Somatic Topology Localization and Temporal Feature Aggregation
- MDSSD: Multi-scale Deconvolutional Single Shot Detector for Small Objects
- CMS-RCNN: Contextual Multi-Scale Region-based CNN for Unconstrained Face Detection 集成人体上下文信息来帮助推理人脸位置；
- Finding Tiny Faces，多级图像金字塔进行multi-scale训练和测试；
- S3FD: Single Shot Scale-invariant Face Detector single shot尺度等变的人脸检测器；
- Repulsion Loss: Detecting Pedestrians in a Crowd
- Deep Direct Regression for Multi-Oriented Scene Text Detection
- TextBoxes: A Fast Text Detector with a Single Deep Neural Network
- Detecting Text in Natural Image with Connectionist Text Proposal Network
- R2CNN: Rotational Region CNN for Orientation Robust Scene Text Detection
- EAST: An Efficient and Accurate Scene Text Detector
- Detecting Oriented Text in Natural Images by Linking Segments
- Arbitrary-Oriented Scene Text Detection via Rotation Proposals
- Scene Text Detection via Holistic, Multi-Channel Prediction
- Deep Matching Prior Network: Toward Tighter Multi-oriented Text Detection
- Solution for Large-Scale Hierarchical Object Detection Datasets with Incomplete Annotation and Data Imbalance
- Comparison Detector: A novel object detection method for small dataset
- Weakly Supervised Deep Detection Networks
- Learning non-maximum suppression
- Improving Object Detection With One Line of Code，soft-nms
- Loss Rank Mining: A General Hard Example Mining Method for Real-time Detectors
目前常用的回归框loss有l2 loss，smooth l1 loss和IoU loss。
- SSD，ssd_understanding
- Faster RCNN，faster_rcnn_understanding
- R-FCN，rfcn_understanding
- ...
- Training Region-based Object Detectors with Online Hard Example Mining，非常有效的针对Regin-based目标检测模型的在线困难样例学习策略。
- COCO
- ...
可视化
# 在tmux或者另一个终端中开启可视化服务器visdom
python -m visdom.server
# 然后在浏览器中查看127.0.0.1:9097训练
# 训练模型
python train.py校验
# 校验模型
python validate.py测试
# 测试模型
python test.py- 实现数据集加载VOC