Part-based R-CNNs for Fine-grained Category Detection(精读)
2014-12-08 17:53
1001 查看
一.文献名字和作者
Part-based R-CNNs for Fine-grained Category Detection, ECCV2014二.阅读时间
2014年12月8日三.文献的目的
文献的目的是解决在进行细粒度识别的测试过程中,关于物体边框注解的依赖问题。四.文献的贡献点
作者提出了一种使用深度卷积特征作为整个物体检测器和可区分性物体检测器,并且在整体和可区分性块上面添加几何约束的细粒度分类算法,该算法在Caltech-UCSD 鸟类数据集上面获得很高的效果。4.1 基于块模型的RCNN
对于每一个类别,训练一个root SVM的w0用于判断该物体是否出现,训练n个part SVM的权值w1,w2,...,wn用于判读块n是否在图像中出现,使用的特征为每一个RCNN学习的深度卷积特征,利用这些SVM可以得到全局和块的得分,当得分大于某个阈值时,可以然后全局或者该块在图像中出现。4.2 几何约束
如果仅仅考虑块检测的话,可能检测出来的块会出现一些超过了物体本身边界的地方(因为在测试的过程中,并没有用到边界,所以才可能出现这种情况)。因此,采用了几何约束,将块检测的结果约束在物体检测的一定范围内。五.使用的数据库
CUB200-2011六.实验结果
版权所有,欢迎转载,转载请注明出处,谢谢
相关文章推荐
- 论文提要“Part-based R-CNNs for Fine-grained Category Detection”
- 论文阅读(3)--SPDA-CNN: Unifying Semantic Part Detection and Abstraction for Fine-grained Recognition
- 论文阅读(4)--Part-Stacked CNN for Fine-Grained Visual Categorization
- SPDA-CNN:Unifying Semantic Part Detection and Abstraction for Fine-grained Recognition
- Fast R-CNN: Fast Region-based Convolutional Networks for object detection(阅读)
- Fast R-CNN: Fast Region-based Convolutional Networks for object detection(阅读)
- BoxCars: 3D Boxes as CNN Input for Improved Fine-Grained Vehicle Recognition
- A Codebook-Free and Annotation-Free Approach for Fine-Grained Image Categorization(精读)
- Unsupervised Template Learning for Fine-Grained Object Recognition(精读)
- Region-based Convolutional Networks for Accurate Object Detection and Segmentation----R-CNN论文笔记
- OpenCV Latent SVM Discriminatively Trained Part Based Models for Object Detection
- 目标检测之LibPaBOD(a LIBrary for PArt-Based Object Detection in C++)
- Region based detection方法的应用: Is Faster R-CNN Doing Well for Pedestrian Detection?
- Fine-Grained Crowdsourcing for Fine-Grained Recognition(精读)
- SPDA-CNN:Unifying Semantic Part Detectiojn and Abstraction for Fine-grained Recognition
- Reading Note: Deformable Part-based Fully Convolutional Network for Object Detection
- LibPaBOD(a LIBrary for PArt-Based Object Detection in C++)我在运行中遇到的问题和解决方案
- Bilinear CNN Models for Fine-grained Visual Recognition
- The Application of Two-level Attention Models in CNN for Fine-grained Image Classification
- OpenCV Latent SVM Discriminatively Trained Part Based Models for Object Detection