Tools for the Reality Technology (實境技術工具介紹)Jian-Kai Wang
本次分享的實境技術主題,將探討虛擬實境(VR)與擴增實境概念(AR)與實現的工具,並說明專屬與非專屬開發平台及開發工具串聯的分工,可以快速掌握實境技術的趨勢與開發方式。
This topic shares you with reality technology, including virtual reality (VR) and augmented reality (AR). This topic also includes the developed tools used in a specific or non-specific platform. Through this topic, you can catch the trend of reality technology.
從圖像辨識到物件偵測,進階的圖影像人工智慧 (From Image Classification to Object Detection, Advance...Jian-Kai Wang
複習及補充機器學習與深度學習,說明物件偵測要解決的問題。
探討策略1: One-Shot Solution,舉 YOLO 為例及其 Hands-On 操作,並探討其他相關演算法與其發展;其次探討策略2: Divide-and-Conquer,以 Faster RCNN 為例與利用 Tensorflow Object Detection API 進行練習,探討其他相關演算法與其發展。
最後探討增進訓練結果與演算法發展,並介紹機器學習的推論與應用與應用機器學習導入產業。
We first reviewed the Machine Learning basis, introduced what object detection is, and then described what the problems it is going to solve. (both the localization and the category issues)
Second, we introduced two types of algorithms that represent two different ideas. One is a One-Shot solution and the other is a divide-and-conquer way. The representative algorithm for the one-shot solution is "YOLO" and the other one is "Faster R-CNN". We also implemented the whole YOLO training and inference processes from scratch via Tensorflow 2.0. On the other hand, we introduced how to use Tensorflow Object Detection APIs to implement the whole Faster R-CNN training and inference processes.
Third, we quickly introduced the evolution of several famous object detection algorithms and how to improve training performance and results.
In the final, we introduced the gap between the AI industrial in research and in practice.
Tools for the Reality Technology (實境技術工具介紹)Jian-Kai Wang
本次分享的實境技術主題,將探討虛擬實境(VR)與擴增實境概念(AR)與實現的工具,並說明專屬與非專屬開發平台及開發工具串聯的分工,可以快速掌握實境技術的趨勢與開發方式。
This topic shares you with reality technology, including virtual reality (VR) and augmented reality (AR). This topic also includes the developed tools used in a specific or non-specific platform. Through this topic, you can catch the trend of reality technology.
從圖像辨識到物件偵測,進階的圖影像人工智慧 (From Image Classification to Object Detection, Advance...Jian-Kai Wang
複習及補充機器學習與深度學習,說明物件偵測要解決的問題。
探討策略1: One-Shot Solution,舉 YOLO 為例及其 Hands-On 操作,並探討其他相關演算法與其發展;其次探討策略2: Divide-and-Conquer,以 Faster RCNN 為例與利用 Tensorflow Object Detection API 進行練習,探討其他相關演算法與其發展。
最後探討增進訓練結果與演算法發展,並介紹機器學習的推論與應用與應用機器學習導入產業。
We first reviewed the Machine Learning basis, introduced what object detection is, and then described what the problems it is going to solve. (both the localization and the category issues)
Second, we introduced two types of algorithms that represent two different ideas. One is a One-Shot solution and the other is a divide-and-conquer way. The representative algorithm for the one-shot solution is "YOLO" and the other one is "Faster R-CNN". We also implemented the whole YOLO training and inference processes from scratch via Tensorflow 2.0. On the other hand, we introduced how to use Tensorflow Object Detection APIs to implement the whole Faster R-CNN training and inference processes.
Third, we quickly introduced the evolution of several famous object detection algorithms and how to improve training performance and results.
In the final, we introduced the gap between the AI industrial in research and in practice.
• 2016-01-26 Presented on group meeting
• [UIST 2015] FoveAR: Combining an Optically See-Through Near-Eye Display with Spatial Augmented Reality Projections
by Hrvoje Benko, Eyal Ofek, Feng Zheng, Andrew D. Wilson
MIT 6.870 - Template Matching and Histograms (Nicolas Pinto, MIT)npinto
MIT 6.870 Object Recognition and Scene Understanding (Fall 2008)
http://people.csail.mit.edu/torralba/courses/6.870/6.870.recognition.htm
This class will review and discuss current approaches to object recognition and scene understanding in computer vision. The course will cover bag of words models, part based models, classifier based models, multiclass object recognition and transfer learning, concurrent recognition and segmentation, context models for object recognition, grammars for scene understanding and large datasets for semi supervised and unsupervised discovery of object and scene categories. We will be reading a mixture of papers from computer vision and influential works from cognitive psychology on object and scene recognition.
Augmented reality for architectural visualisation Gulnaz Aksenova
Workshop made on the use of Augmented reality for architectural information visualization at the summer school of architecture and urbanism "Sochi-Peshkom" in city Sochi, Russia
Virtual Reality (VR) Continuum - AMP New VenturesAMP New Ventures
If the Internet is the sharing of information, then Virtual Reality (VR) is the sharing of experiences; and if most customer experiences are digital, then Virtual Reality (VR) must be important, for it is the next frontier in digital.
VR immerses users in indistinguishably real simulated environments, while Augmented Reality (AR) blends the digital into our physical environments. In the past month, PlayStation VR was released along with Google VR, to join a global ecosystem of VR content, infrastructure and platforms startups, projected to be worth $160bn by 2020.
Given It will transform experiences across industries, including Financial Services, and the expert consensus is that mainstream adoption is ~5 years away, we recommend Financial Services companies start exploring VR/AR possibilities now.
Biomechanical, symbiotic communication between the virtual and physical worlds. Understanding biosyncing is crucial to designing and building products that rely on the interaction between humans and machines.
• 2016-01-26 Presented on group meeting
• [UIST 2015] FoveAR: Combining an Optically See-Through Near-Eye Display with Spatial Augmented Reality Projections
by Hrvoje Benko, Eyal Ofek, Feng Zheng, Andrew D. Wilson
MIT 6.870 - Template Matching and Histograms (Nicolas Pinto, MIT)npinto
MIT 6.870 Object Recognition and Scene Understanding (Fall 2008)
http://people.csail.mit.edu/torralba/courses/6.870/6.870.recognition.htm
This class will review and discuss current approaches to object recognition and scene understanding in computer vision. The course will cover bag of words models, part based models, classifier based models, multiclass object recognition and transfer learning, concurrent recognition and segmentation, context models for object recognition, grammars for scene understanding and large datasets for semi supervised and unsupervised discovery of object and scene categories. We will be reading a mixture of papers from computer vision and influential works from cognitive psychology on object and scene recognition.
Augmented reality for architectural visualisation Gulnaz Aksenova
Workshop made on the use of Augmented reality for architectural information visualization at the summer school of architecture and urbanism "Sochi-Peshkom" in city Sochi, Russia
Virtual Reality (VR) Continuum - AMP New VenturesAMP New Ventures
If the Internet is the sharing of information, then Virtual Reality (VR) is the sharing of experiences; and if most customer experiences are digital, then Virtual Reality (VR) must be important, for it is the next frontier in digital.
VR immerses users in indistinguishably real simulated environments, while Augmented Reality (AR) blends the digital into our physical environments. In the past month, PlayStation VR was released along with Google VR, to join a global ecosystem of VR content, infrastructure and platforms startups, projected to be worth $160bn by 2020.
Given It will transform experiences across industries, including Financial Services, and the expert consensus is that mainstream adoption is ~5 years away, we recommend Financial Services companies start exploring VR/AR possibilities now.
Biomechanical, symbiotic communication between the virtual and physical worlds. Understanding biosyncing is crucial to designing and building products that rely on the interaction between humans and machines.
The Gandhi Heritage Portal is a repository of authentic information about Mahatma Gandhi. It contains rare photos, books and writings about Mahatma Gandhi.
MasTool, bankaların süreçlerinin Bankacılık Denetleme ve Düzenleme Kurumu (BDDK) yönetmelik ve genelgelerine uygun olarak hazırlanmasını sağlar. Süreçlerin Risk Kontrol Matrislerine (RCM) göre denetimini kolaylaştırır. Bu kapsamda yapılan test sonuçlarının girilmesini, bulguların sorumlu kişilerle / gruplarla paylaşılmasını, cevap / aksiyon planlarının alınmasını ve takibini, yıllık olarak Süreç Sahibi Beyan Mektupları alınarak Yönetim Beyan Mektubunun hazırlanmasını sağlayan bir üründür.
Detaylı bilgi için tıklayınız: http://mirsis.com.tr/Mastool
11. 數位學習的新趨勢 圖片摘錄自 : http://www.i-tlog.com/lemonadeplease/entry/2010 數位學 習 科技趨勢分析 08/16/11 National Defense University - Department of Computer Science.
12.
13. 何謂擴增實境 在 1994 年 Milgram 將真實 場景與 虛擬 實境 之間的變化程度, 提出了「真實 - 虛擬連續性 (reality–virtuality Continuum) 」 理 論 ,為擴增實境與虛擬實境做了有所區別。 圖片摘錄自: Michael Haller el. ,“ Emerging Technologies of Augmented Reality : Interfaces and Design” , Idea Group Publishing , U.S.A , 2007 08/16/11 National Defense University - Department of Computer Science.
14.
15. 擴增實境 運作步驟 08/16/11 National Defense University - Department of Computer Science. 載入 追蹤辨識 定位疊加 顯示
16. 擴增實境之類別 標記 (Marker) 08/16/11 National Defense University - Department of Computer Science. 無標記 (Markerless)
17.
18.
19.
20.
21.
22. 現有開發平台分析 08/16/11 National Defense University - Department of Computer Science. ARToolKit D’Fusion Unifeye Design
41. 移動物件偵測—光流法 08/16/11 National Defense University - Department of Computer Science. 光流法 (Optical Flow Method) 的原理,簡單 的說 ,光流法就是 藉由偵測光線強弱的變化來進行偵測移動物件。
42. 移動物件偵測—時序差異法 08/16/11 National Defense University - Department of Computer Science. 影像 K-1 影像 K 物件變動位置 差異影像 時序差異法 (Temporal Differencing )的原理, 簡單 的說 利用連續 2 幅 至 3 幅的影像做 1 對 1 的像素相減 , 若 兩者差異不為零則為 移動 物件。
43. 移動物件偵測—背景相減法 National Defense University - Department of Computer Science. 08/16/11 背景相減法 (Background Subtraction) 的原理 : 1. 運用幾秒的連續影像建立初步背景 2. 在以目前影像與背景影像比較, 3. 符合條件為背景,反之為 移動 物件。
44.
45.
46.
47. 座標轉換 08/16/11 National Defense University - Department of Computer Science. 其中 K 為相機校正矩陣公式 : C x 與 C y 分別代表 x 與 y 上的偏移量 。 f x 與 f y 分別代表 x 與 y 含比例參數的座標點 。 物理意義上的像素也不一定是矩形, s 則代表 X 軸與 Y 軸的歪斜參數。 轉換公式: Q 為世界座標系統三維座標點,而世界座標系統三維座標點與相機座標系統相對三維座標點的關係,可由位移轉換 t 與旋轉轉換 R 來表示。 公式中: