2017CVPRSelf-Supervised Learning of Visual Features Through Embedding Images Into Text Topic Spaces
這個工作的背景是無監督學習，也就是現存的網絡上有大量數據，如何在不使用人工進行標記的情況下來進行特徵提取的學習，是一個未來很有前景的方向，比如網絡上無論是社交網站還是wiki百科，都有大量圖文並茂的資源，作者的意思就是利用這些」文「來指導「圖」的特徵提取，不是一般意義上的無監督學習。作者利用自己訓練的網絡當作特徵提取器在分類，目標檢測和圖文互檢索的任務上進行測試，有大量試驗，但作者的效果並不是相比
相關文章
- 1. [解讀] Deep Clustering for Unsupervised Learning of Visual Features
- 2. Deep Clustering for Unspervised Learning of Visual Features
- 3. Learning like a Child: Fast Novel Visual Concept Learning from Sentence Descriptions of Images
- 4. Inspirational Applications of Deep Learning
- 5. 閱讀筆記：VISUALCOMFORTASSESSMENTOFSTEREOSCOPICIMAGES USINGDEEPVISUALANDDISPARITYFEATURESBAS
- 6. 【論文筆記】Embedding of Embedding (EOE) : Joint Embedding for Coupled Heterogeneous Networks
- 7. A NEW METHOD OF REGION EMBEDDING FOR TEXT CLASSIFICATION
- 8. Overview of Zero-Shot Learning - 1.3
- 9. paper review : Deep Audio-Visual learning: A Survey
- 10. Hands-On Learning Through Racing
- 更多相關文章...
- • Docker images 命令 - Docker命令大全
- • SVG
- SVG 教程 - • Docker 清理命令
- • JDK13 GA發佈：5大特性解讀
相關標籤/搜索
每日一句
-
每一个你不满意的现在，都有一个你没有努力的曾经。
最新文章
歡迎關注本站公眾號,獲取更多信息
相關文章
- 1. [解讀] Deep Clustering for Unsupervised Learning of Visual Features
- 2. Deep Clustering for Unspervised Learning of Visual Features
- 3. Learning like a Child: Fast Novel Visual Concept Learning from Sentence Descriptions of Images
- 4. Inspirational Applications of Deep Learning
- 5. 閱讀筆記：VISUALCOMFORTASSESSMENTOFSTEREOSCOPICIMAGES USINGDEEPVISUALANDDISPARITYFEATURESBAS
- 6. 【論文筆記】Embedding of Embedding (EOE) : Joint Embedding for Coupled Heterogeneous Networks
- 7. A NEW METHOD OF REGION EMBEDDING FOR TEXT CLASSIFICATION
- 8. Overview of Zero-Shot Learning - 1.3
- 9. paper review : Deep Audio-Visual learning: A Survey
- 10. Hands-On Learning Through Racing