Pedestrian-Synthesis-GAN: Generating Pedestrian Data in Real Scene and Beyond
Abstract. 最早進的行人檢測模型在許多基準測試中取得了巨大的成功。然而，這些模型須要大量的註釋信息，標記過程一般須要花費大量的時間和精力。本文提出了一種生成標記行人數據的方法，並將其用於行人檢測器的訓練。該框架創建在具備多個鑑別器的生成對抗網絡(GAN)的基礎上，試圖對真實行人進行綜合，同時學習背景信息。爲了處理不一樣大小的行人，咱們在鑑別器中採用了空間金字塔池(SPP)層。咱們在兩個基準
相關文章
- 1. Display real-time data in Angular
- 2. Generating Videos With Scene Dynamics
- 3. WebRTC in the real world: STUN, TURN and signaling
- 4. A real example of vioplot in R (sample data and code attached)
- 5. Real-Time Pedestrian Detection With Deep Network Cascades
- 6. Research on Pedestrian Attribute Recognition Based on Semantic Segmentation in Natural Scene
- 7. Opentracing and Jaeger in a real Go Microservices
- 8. AOD-NET：An All-in-One Network for Dehazing and Beyond
- 9. Real-Time Pedestrian Detection with Deep Network Cascades
- 10. 人羣運動--Scene-Independent Group Profiling in Crowd
- 更多相關文章...
- • SQL IN 操作符 - SQL 教程
- • Swift for-in 循環 - Swift 教程
- • RxJava操作符（七）Conditional and Boolean
- • 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。
歡迎關注本站公眾號,獲取更多信息
相關文章
- 1. Display real-time data in Angular
- 2. Generating Videos With Scene Dynamics
- 3. WebRTC in the real world: STUN, TURN and signaling
- 4. A real example of vioplot in R (sample data and code attached)
- 5. Real-Time Pedestrian Detection With Deep Network Cascades
- 6. Research on Pedestrian Attribute Recognition Based on Semantic Segmentation in Natural Scene
- 7. Opentracing and Jaeger in a real Go Microservices
- 8. AOD-NET：An All-in-One Network for Dehazing and Beyond
- 9. Real-Time Pedestrian Detection with Deep Network Cascades
- 10. 人羣運動--Scene-Independent Group Profiling in Crowd