Fast R-CNN
By Ross B. Girshick. 2015
28,747 citations recorded by the source corpus.
This paper proposes Fast R-CNN, a clean and fast framework for object detection. Compared to traditional R-CNN, and its accelerated version SPPnet, Fast R-CNN trains networks using a multi-task loss in a single training stage. The multi-task loss simplifies learning and improves detection accuracy. Unlike SPPnet, all network layers can be updated during fine-tuning. We show that this difference has practical ramifications for very deep networks, such as VGG16, where mAP suffers when only the fully-connected layers are updated. Compared to"slow"R-CNN, Fast R-CNN is 9x faster at training VGG16 for detection, 213x faster at test-time, and achieves a significantly higher mAP on PASCAL VOC 2012. Compared to SPPnet, Fast R-CNN trains VGG16 3x faster, tests 10x faster, and is more accurate. Fast R-CNN is implemented in Python and C++ and is available under the open-source MIT License at https://github.com/rbgirshick/fast-rcnn
Subfields: Computer Vision and Pattern Recognition, Neurology
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Computer Vision and Pattern Recognition
- Luc Van Gool
- Dacheng Tao
- Xiaogang Wang
- Ming-Hsuan Yang
- Trevor Darrell
- Jitendra Malik
- Yi Yang
- Andrew Zisserman
In Neurology
- Andrea Vedaldi
- Paolo Favaro
- Song Han
- Abhinav Kumar Gupta
- Ishan Misra
- Chunhua Shen
- Kaiming He
- Bharath Hariharan
Citations in the corpus
- Deep Residual Learning for Image Recognition (2015, background): “Many other nontrivial visual recognition tasks [8, 12, 7, 32, 27] have also”
- Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks (2015, methodology): “, Selective Search [4] object detectors, RCNN [5], and Fast R-CNN [2]).”
- Object Detection via a Multi-region and Semantic Segmentation-Aware CNN Model (2015, methodology): “Contemporary to our work are the approaches of [29, 9, 28] that are also based on the SPP-Net framework.”
- Object Detection Networks on Convolutional Feature Maps (2015, methodology): “plored families in previous works [12], [13], [14].”
- Delving Deep into Rectifiers: Surpassing Human-Level Performance on ImageNet Classification (2015, methodology): “Object detectionmAP on PASCALVOC 2007 using Fast R-CNN [6] on different pre-trained nets.”
- Mask R-CNN (2017, methodology): “The classification loss Lcls and bounding-box loss Lbox are identical as those defined in [12].”
- Momentum Contrast for Unsupervised Visual Representation Learning (2019, methodology): “ImageNet supervised pre-training is most influential when serving as the initialization for fine-tuning in downstream tasks ( e.g ., [21, 20, 43, 52]).”
- Rethinking ImageNet Pre-Training (2018, background): “More surprisingly, we can achieve these results by using baseline systems [8, 36, 26, 13] and their hyper-parameters that were optimized for fine-tuning pre-trained models.”
- An Overview of Multi-Task Learning in Deep Neural Networks (2017, methodology): “Multi-task learning has been used successfully across all applications of machine learning, from natural language processing [Collobert and Weston, 2008] and speech recognition [Deng et al., 2013] to computer vision [Girshick, 2015] and drug discovery [Ramsundar et al., 2015].”
- Growing a Brain: Fine-Tuning by Increasing Model Capacity (2017, methodology): “Also, following the standard practice in Fast R-CNN [9], we fine-tune from the layer C onv 2 1 in the “All” scenario.”
- Exploring Self-Attention for Image Recognition (2020, background): “These serve as the basis for models developed for other computer vision tasks, such as semantic segmentation [22, 3, 42, 44] and object detection [10, 9, 28, 21].”
- Group Normalization (2018, methodology): “For example, the Fast/er and Mask R-CNN frameworks [12, 47, 18] use a batch size of 1 or 2 images because of higher resolution, where BN is “frozen” by transforming to a linear layer [20]; in video classification with 3D convolutions [60, 6], the presence of spatial-temporal features introduces a tr”
- Relation Networks for Object Detection (2017, methodology): “The state-of-the-art object detection methods [24, 18, 38, 9, 32, 10, 23] mostly follow the region based paradigm since it is established in the seminal work R-CNN [19].”
- Self-Supervised Learning of Visual Features through Embedding Images into Text Topic Spaces (2017, positive): “For object detection we fine-tune our classification network using Fast R-CNN [12] with default parameters for 40, 000 iterations.”
- Visual Genome: Connecting Language and Vision Using Crowdsourced Dense Image Annotations (2016, background): “Object detection is already a well-studied problem (Everingham et al. 2010; Girshick et al. 2014; Sermanet et al. 2013; Girshick 2015; Ren et al. 2015b).”
- Unsupervised Feature Learning via Non-parametric Instance Discrimination (2018, methodology): “We experiment with Fast R-CNN [7] with AlexNet and VGG16 architectures, and Faster R-CNN [32] with ResNet-50.”
- Unsupervised Representation Learning by Predicting Image Rotations (2018, background): “For instance, the image features learned by ConvNets in this supervised manner have achieved excellent results when they are transferred to other vision tasks, such as object detection (Girshick, 2015), semantic segmentation (Long et al., 2015), or image captioning (Karpathy & Fei-Fei, 2015).”
- Accurate, Large Minibatch SGD: Training ImageNet in 1 Hour (2017, methodology): “As an extension of the image-centric Fast/Faster R-CNN [9, 30], Mask R-CNN exhibits different minibatch sizes for different layers : the network backbone uses two images (per GPU), but each image contributes 512 Regions-of-Interest for computing classification (multinomial cross-entropy),…”
- Learning Features by Watching Objects Move (2016, methodology): “We first evaluate our representation on the task of object detection using Fast R-CNN.”
- Feature Pyramid Networks for Object Detection (2016, methodology): “Inference time increases considerably ( e.g ., by four times [11]), making this approach impractical for real applications.”