Face R-CNN論文解讀
網路結構:
這篇文章在基於faster rcnn做了一些改進:
First, based on a newly developed loss function called center loss [33], we design a new multi-task
loss function in the Fast R-CNN model to supervise the learning of discriminative deep features for
face/non-face classification.
Second, in the learning of deep features, we use online hard example
mining algorithm [1] to generate hard samples (in which the ratio of positive samples to negative
samples is set to be 1:1) for subsequent processing.
Third, we use multi-scale training strategy to
help improve the detection performance.
這篇文章與faster rcnn的不同之處主要在fast rcnn部分
1、Center Loss
fast rcnn的損失函式變為:
為了使得center loss均衡,一個mini batch中正負樣本比例限制為1:1.
2、Online Hard Example Mining
每次從正負樣本中各選出loss最大的N個樣本加入下次訓練。
3、Multi-Scale Training
Instead of using a fixed scale for all the training images in the typical Faster R-CNN framework, we
design a multi-scale representation for each image by resizing the original image to different sizes
during the training process. In this way, the learned model is more adapted to low-resolution faces.
In the testing process, multi-scale testing is performed accordingly, and the predicted bounding
boxes at different image scales are combined into the final output.
為了弱化尺度影響(或者更好地檢測小目標),訓練階段圖片會經過不同尺度縮放。
We train the detector using VGG19 with the ImageNet pre-trained model.
相關文章
- Sparse R-CNN: End-to-End Object Detection with Learnable Proposals 論文解讀CNNObject
- 論文提要 Deep Face Recognition
- 《Stereo R-CNN based 3D Object Detection for Autonomous Driving》論文解讀CNN3DObject
- Reading Face, Read Health論文閱讀筆記筆記
- Face++ 論文解讀:一種新的行人重識別度量學習方法
- 【論文研讀】Recurrent convolutional strategies for face manipulation detection in videosIDE
- 近期有哪些值得讀的QA論文?| 專題論文解讀
- Sparse R-CNN: End-to-End Object Detection with Learnable Proposals - 論文閱讀翻譯CNNObject
- NCSU&阿里巴巴論文:可解釋的R-CNN阿里CNN
- PointNet系列論文解讀
- DeepSort論文解讀
- EfficientNet & EfficientDet 論文解讀
- Light Head R-CNN論文筆記CNN筆記
- 一文讀懂目標檢測:R-CNN、Fast R-CNN、Faster R-CNN、YOLO、SSDCNNASTYOLO
- 論文解讀(DGI)《DEEP GRAPH INFOMAX》
- 論文解讀《Cauchy Graph Embedding》
- Backbone 網路-DenseNet 論文解讀SENet
- Backbone 網路-ResNet 論文解讀
- 深度學習論文翻譯解析(十二):Fast R-CNN深度學習ASTCNN
- 論文解讀|李國良等:Database Meets AIDatabaseAI
- 論文解讀(GAN)《Generative Adversarial Networks》
- [論文解讀]Baidu Apollo EM Motion PlannerAI
- 論文解讀丨表格識別模型TableMaster模型AST
- 怎樣讀論文?
- 論文閱讀:SiameseFC
- 論文解讀(BGRL)《Bootstrapped Representation Learning on Graphs》bootAPP
- 論文解讀(GCC)《Graph Contrastive Clustering》GCAST
- 論文解讀(IGSD)《Iterative Graph Self-Distillation》
- 論文解讀(Debiased)《Debiased Contrastive Learning》AST
- 論文解讀SDCN《Structural Deep Clustering Network》Struct
- [論文解讀]A Quantitative Analysis Framework for Recurrent Neural NetworkFramework
- 論文解讀(SDNE)《Structural Deep Network Embedding》Struct
- 論文解讀(GIN)《How Powerful are Graph Neural Networks》
- 論文解讀(DFCN)《Deep Fusion Clustering Network》
- 端到端文字識別CRNN論文解讀RNN
- 論文解讀(LLE)《Nonlinear Dimensionality Reduction by Locally Linear Embedding》以及論文通俗解釋
- 如何閱讀科研論文
- 並行多工學習論文閱讀(五):論文閱讀總結並行