【導讀】計算機視覺領域的頂級會議ECCV2018於9月8日在德國慕尼黑舉辦,前兩天是workshop日程。在主會議正式開幕之前,讓我們先來看看24位ECCV2018論文作者開源的論文實現代碼~
1. Simple Baselines for Human Pose Estimation and Tracking (357 stras)
論文連結:https://arxiv.org/abs/1804.06208
代碼連結:https://github.com/Microsoft/human-pose-estimation.pytorch#simple-baselines-for-human-pose-estimation-and-tracking
2. ICNet for Real-Time SemanticSegmentation on High-Resolution Images (344 stars)
論文連結:https://hszhao.github.io/projects/icnet/
代碼連結:https://github.com/hszhao/ICNet
3. Instance-Batch Normalization Network (225stars)
論文連結:https://arxiv.org/abs/1807.09441
代碼連結:https://github.com/XingangPan/IBN-Net
4. Distractor-aware SiameseNetworks for Visual Object Tracking (225 stars)
論文連結:https://arxiv.org/pdf/1808.06048.pdf
代碼連結:https://github.com/foolwood/DaSiamRPN
5. Pixel2Mesh: Generating3D Mesh Models from Single RGB Images (98 stars)
論文連結:https://arxiv.org/abs/1804.01654
代碼連結:https://github.com/nywang16/Pixel2Mesh
6. MVSNet: Depth Inferencefor Unstructured Multi-view Stereo(85 stars)
論文連結:https://arxiv.org/abs/1804.02505
代碼連結:https://github.com/YoYo000/MVSNet
7. Macro-Micro AdversarialNetwork for Human Parsing (61stars)
論文連結:https://arxiv.org/abs/1807.08260
代碼連結:https://github.com/RoyalVane/MMAN
8. Learning Human-Object Interactions by GraphParsing Neural Networks (54 stars)
論文連結:http://web.cs.ucla.edu/~syqi/publications/eccv2018gpnn/eccv2018gpnn.pdf
代碼連結:https://github.com/SiyuanQi/gpnn
9. Enhanced Super-Resolution GenerativeAdversarial Networks
論文連結:https://arxiv.org/abs/1809.00219
代碼連結:https://github.com/xinntao/ESRGAN
10. PSANet: Point-wise Spatial AttentionNetwork for Scene Parsing (34 stars)
論文連結:https://hszhao.github.io/projects/psanet/
代碼連結:https://github.com/hszhao/PSANet
11.OM-CNN+2C-LSTM for video salinecy prediction
論文連結:https://arxiv.org/abs/1709.06316
代碼連結:https://github.com/remega/OMCNN_2CLSTM
12. Interpretable Intuitive Physics Model
論文連結:https://www.cs.cmu.edu/~xiaolonw/papers/ECCV_Physics_Cameraready.pdf
代碼連結:https://github.com/tianye95/interpretable-intuitive-physics-model
13. Transferring GANs generating images fromlimited data
論文連結:https://arxiv.org/abs/1805.01677
代碼連結:https://github.com/yaxingwang/Transferring-GANs
14. Superpixel Sampling Networks
論文連結:https://varunjampani.github.io/ssn/
代碼連結:https://github.com/NVlabs/ssn_superpixels
15. A Trilateral Weighted Sparse Coding Schemefor Real-World Image Denoising
論文連結:https://arxiv.org/abs/1807.04364
代碼連結:https://github.com/csjunxu/TWSC-ECCV2018
16. Deep Randomized Ensembles for MetricLearning
論文連結:https://arxiv.org/abs/1808.04469
代碼連結:https://github.com/littleredxh/DREML
17. VISDrone2018: Challenge-ObjectDetection in Images
競賽內容主頁:http://www.aiskyeye.com/
實現代碼:https://github.com/zhpmatrix/VisDrone2018
18. Part-Aligned Bilinear Representations forPerson Re-identification
論文連結:https://cv.snu.ac.kr/publication/conf/2018/reid_eccv18.pdf
代碼連結:https://github.com/yuminsuh/part_bilinear_reid
19. Inner Space Preserving - Generative PoseMachine (ISP-GPM)
論文連結:https://arxiv.org/abs/1808.02104
代碼連結:https://github.com/ostadabbas/isp-gpm
20. Rethinking the Form of Latent States inImage Captioning
論文連結:https://arxiv.org/abs/1807.09958
代碼連結:https://github.com/doubledaibo/2dcaption_eccv2018
21. Diverse Conditional Image Generation byStochastic Regression with Latent Drop-Out Codes
論文連結:https://arxiv.org/abs/1808.01121
代碼連結:https://github.com/SSAW14/Image_Generation_with_Latent_Code
22. Learning Efficient Single-stage PedestrianDetectors by Asymptotic Localization Fitting
論文連結:https://github.com/liuwei16/ALFNet/blob/master/docs/2018ECCV-ALFNet.pdf
代碼連結:https://github.com/liuwei16/ALFNet
23. Statistically-motivated Second-orderPooling
論文連結:https://arxiv.org/abs/1801.07492
代碼連結:https://github.com/kcyu2014/smsop
24. Learning to Navigate for Fine-grainedClassification
論文連結:https://arxiv.org/abs/1809.00287
代碼連結:https://github.com/yangze0930/NTS-Net
-END-
專 · 知
人工智慧領域26個主題知識資料全集獲取與加入專知人工智慧服務群: 歡迎微信掃一掃加入專知人工智慧知識星球群,獲取專業知識教程視頻資料和與專家交流諮詢!
請PC登錄www.zhuanzhi.ai或者點擊閱讀原文,註冊登錄專知,獲取更多AI知識資料!
請加專知小助手微信(掃一掃如下二維碼添加),加入專知主題群(請備註主題類型:AI、NLP、CV、 KG等)交流~
AI 項目技術 & 商務合作:bd@zhuanzhi.ai, 或掃描上面二維碼聯繫!請關注專知公眾號,獲取人工智慧的專業知識!
點擊「閱讀原文」,使用專知