Proposes a deep neural image captioning model combining CNN and LSTM, achieving BLEU-4 of 27.7, outperforming previous methods.

Oriol Vinyals, Alexander Toshev, Samy Bengio et al.

Proposes LRCN, combining CNN and LSTM for video recognition and captioning, outperforming single models with end-to-end training.

Jeff Donahue, Lisa Anne Hendricks, Marcus Rohrbach et al.

Proposes Fully Convolutional Networks (FCN) for pixel-level semantic segmentation, achieving 62.2% mean IU, surpassing previous SOTA with end-to-end training.

Jonathan Long, Evan Shelhamer, Trevor Darrell

Dictionary learning combined with `1-norm sparse coding improves image representation and recognition accuracy.

Julien Mairal, Francis Bach, Jean Ponce

Proposed Inception architecture employs multi-scale convolutions and dimension reduction, achieving 28.6% Top-5 error on ImageNet with 1/12 parameters of AlexNet.

Christian Szegedy, Wei Liu, Yangqing Jia et al.

Deep CNNs like AlexNet, VGG, ResNet trained on 14 million images achieved top-5 error rates below 7%, revolutionizing large-scale image recognition.

Olga Russakovsky, Jia Deng, Hao Su et al.

MS COCO dataset with 91 categories, 2.5 million instances, enhances scene understanding and precise localization.

Tsung-Yi Lin, Michael Maire, Serge Belongie et al.

Exploiting linear filter structures with low-rank SVD and clustering speeds up CNN evaluation 2-3×, with <1% accuracy loss.

Remi Denton, Wojciech Zaremba, Joan Bruna et al.

Proposes multi-sentence, multi-level video description framework using semantic representations and visual recognition, improving coherence and detail control.

Anna Senina, Marcus Rohrbach, Wei Qiu et al.

Using pretrained OverFeat features with linear SVM achieves state-of-the-art results across multiple vision tasks without fine-tuning.

Ali Sharif Razavian, Hossein Azizpour, Josephine Sullivan et al.

Introduces multi-scale invertible orientation scores for vessel enhancement, significantly improving crossing and bifurcation detection in retinal images.

Julius Hannink, Remco Duits, Erik Bekkers

Gradient-based visualization of CNN class models and saliency maps enhances interpretability and weakly supervised object localization.

Karen Simonyan, Andrea Vedaldi, Andrew Zisserman

Proposes FFT-based convolution acceleration, achieving over 10x speedup in training and inference for deep CNNs.

Michael Mathieu, Mikael Henaff, Yann LeCun

DeCAF features extracted from ImageNet-trained CNN outperform traditional features on multiple vision tasks, achieving over 20% accuracy improvements.

Jeff Donahue, Yangqing Jia, Oriol Vinyals et al.

Introduces an efficient inference algorithm for fully connected CRFs, enhancing image segmentation accuracy.

Philipp Krähenbühl, Vladlen Koltun