Search for: All records

Face recognition in collaborative learning videos presents many challenges. In collaborative learning videos, students sit around a typical table at different positions to the recording camera, come and go, move around, get partially or fully occluded. Furthermore, the videos tend to be very long, requiring the development of fast and accurate methods. We develop a dynamic system of recognizing participants in collaborative learning systems. We address occlusion and recognition failures by using past information about the face detection history. We address the need for detecting faces from different poses and the need for speed by associating each participant with a collection of prototype faces computed through sampling or K-means clustering. Our results show that the proposed system is proven to be very fast and accurate. We also compare our system against a baseline system that uses InsightFace [2] and the original training video segments. We achieved an average accuracy of 86.2% compared to 70.8% for the baseline system. On average, our recognition rate was 28.1 times faster than the baseline system. 

Long-term object detection requires the integration of frame-based results over several seconds. For non-deformable objects, long-term detection is often addressed using object detection followed by video tracking. Unfortunately, tracking is inapplicable to objects that undergo dramatic changes in appearance from frame to frame. As a related example, we study hand detection over long video recordings in collaborative learning environments. More specifically, we develop long-term hand detection methods that can deal with partial occlusions and dramatic changes in appearance. Our approach integrates object-detection, followed by time projections, clustering, and small region removal to provide effective hand detection over long videos. The hand detector achieved average precision (AP) of 72% at 0.5 intersection over union (IoU). The detection results were improved to 81% by using our optimized approach for data augmentation. The method runs at 4.7× the real-time with AP of 81% at 0.5 intersection over the union. Our method reduced the number of false-positive hand detections by 80% by improving IoU ratios from 0.2 to 0.5. The overall hand detection system runs at 4× real-time. 

Speech recognition is very challenging in student learning environments that are characterized by significant cross-talk and background noise. To address this problem, we present a bilingual speech recognition system that uses an interactive video analysis system to estimate the 3D speaker geometry for realistic audio simulations. We demonstrate the use of our system in generating a complex audio dataset that contains significant cross-talk and background noise that approximate real-life classroom recordings. We then test our proposed system with real-life recordings. In terms of the distance of the speakers from the microphone, our interactive video analysis system obtained a better average error rate of 10.83% compared to 33.12% for a baseline approach. Our proposed system gave an accuracy of 27.92% that is 1.5% better than Google Speech-to-text on the same dataset. In terms of 9 important keywords, our approach gave an average sensitivity of 38% compared to 24% for Google Speech-to-text, while both methods maintained high average specificity of 90% and 92%. On average, sensitivity improved from 24% to 38% for our proposed approach. On the other hand, specificity remained high for both methods (90% to 92%). 

We study the problem of detecting talking activities in collaborative learning videos. Our approach uses head detection and projections of the log-magnitude of optical flow vectors to reduce the problem to a simple classification of small projection images without the need for training complex, 3-D activity classification systems. The small projection images are then easily classified using a simple majority vote of standard classifiers. For talking detection, our proposed approach is shown to significantly outperform single activity systems. We have an overall accuracy of 59% compared to 42% for Temporal Segment Network (TSN) and 45% for Convolutional 3D (C3D). In addition, our method is able to detect multiple talking instances from multiple speakers, while also detecting the speakers themselves. 

We introduce the problem of detecting a group of students from classroom videos. The problem requires the detection of students from different angles and the separation of the group from other groups in long videos (one to one and a half hours). We use multiple image representations to solve the problem. We use FM components to separate each group from background groups, AM-FM components for detecting the back-of-the-head, and YOLO for face detection. We use classroom videos from four different groups to validate our approach. Our use of multiple representations is shown to be significantly more accurate than the use of YOLO alone. 

Research on video activity recognition has been primarily focused on differentiating among many diverse activities defined using short video clips. In this paper, we introduce the problem of reliable video activity recognition over long videos to quantify student participation in collaborative learning environments (45 minutes to 2 hours). Video activity recognition in collaborative learning environments contains several unique challenges. We introduce participation maps that identify how and when each student performs each activity to quantify student participation. We present a family of low-parameter 3D ConvNet architectures to detect these activities. We then apply spatial clustering to identify each participant and generate student participation maps using the resulting detections. We demonstrate the effectiveness by training over about 1,000 3-second samples of typing and writing and test our results over ten video sessions of about 10 hours. In terms of activity detection, our methods achieve 80% accuracy for writing and typing that match the recognition performance of TSN, SlowFast, Slowonly, and I3D trained over the same dataset while using 1200x to 1500x fewer parameters. Beyond traditional video activity recognition methods, our video activity participation maps identify how each student participates within each group. 

Large scale training of Deep Learning methods requires significant computational resources. The use of transfer learning methods tends to speed up learning while producing complex networks that are very hard to interpret. This paper investigates the use of a low-complexity image processing system to investigate the advantages of using AM-FM representations versus raw images for face detection. Thus, instead of raw images, we consider the advantages of using AM, FM, or AM-FM representations derived from a low-complexity filterbank and processed through a reduced LeNet-5. The results showed that there are significant advantages associated with the use of FM representations. FM images enabled very fast training over a few epochs while neither IA nor raw images produced any meaningful training for such low-complexity network. Furthermore, the use of FM images was 7x to 11x faster to train per epoch while using 123x less parameters than a reduced-complexity MobileNetV2, at comparable performance (AUC of 0.79 vs 0.80). 

The dominant use of Convolutional Neural Networks (CNNs) in several image and video analysis tasks necessitates a careful re-evaluation of the underlying software libraries for computing them for large-scale image and video databases. We focus our attention on developing methods that can be applied to large image databases or videos of large image sizes. We develop a method that maximizes throughput through the use of vector-based memory I/O and optimized 2D FFT libraries that run on all available physical cores. We also show how to decompose arbitrarily large images into smaller, optimal blocks that can be effectively processed through the use of overlap-and- add. Our approach outperforms Tensorflow for 5x5 kernels and significantly outperforms Tensorflow for 11x11 kernels. 

We introduce a new method to detect student group interactions in collaborative learning videos. We consider the following video activities: (i) human to human, (ii) human to others, and (iii) lack of any interaction. The system uses multidimensional AM-FM methods to detect student faces, hair, and then use the results to detect possible interactions. We use dynamic graphs to represent group interactions within each video.We tested our methods with 15 videos and achieved an 84% accuracy for students facing the camera and 76% for students facing both towards and away from the camera.