Decoding Children's Social Behavior

Question

1. What are the contributions mentioned in the paper "Decoding children’s social behavior" ?

2. How can the authors obtain information about the state of the dyad?

3. What is the contribution of this paper?

4. What is the main source of error in the tracker?

Accepted Answer

The authors introduce a new problem domain for activity recognition: the analysis of children ’ s social and communicative behaviors based on video and audio data.. The authors introduce a new publicly-available dataset containing over 160 sessions of a 3–5 minute child-adult interaction.. In each session, the adult examiner followed a semistructured play interaction protocol which was designed to elicit a broad range of social behaviors.. The authors identify the key technical challenges in analyzing these behaviors, and describe methods for decoding the interactions.. The authors present experimental results that demonstrate the potential of the dataset to drive interesting research questions, and show preliminary results for multi-modal activity recognition.

Accepted Answer

By leveraging the statistical regularities of adult speech (and other modalities), the authors can obtain valuable information about the state of the dyad.

Accepted Answer

The contribution of this paper is the introduction of the Multimodal Dyadic Behavior (MMDB) dataset which contains this interaction data, along with an initial series of single mode and multimodal analyses to segment, classify and measure relevant behaviors across numerous play interactions.

Accepted Answer

The tracker keeps an ensemble of appearance templates for each target and uses the tracker hierarchy to automatically handle tracker initialization, termination and tracking failure.

Accepted Answer

To date, 121 children between the ages of 15 and 30 months have participated in the Rapid-ABC assessment, and their parents have consented to share their recorded data with the research community.

Accepted Answer

The authors empirically selected 20× 20 bins for the 2D histogram and pyramid level as [1, 2] for all their experiments, yielding a 1200-dimensional feature for each substage.

Accepted Answer

Note that in 9 out of 70 stages, the Nexidia tool generated false positives, detecting phrases more than 5 seconds after the true start of the stage.

Accepted Answer

The authors performed their analysis on 20 hand-picked sessions in which the child remained at the table throughout and the tracker worked successfully.

Accepted Answer

The authors employed commercial software from Omron, the OKAO Vision Library, to detect and track the child’s face, and obtain measures of face detection confidence, smile degree (the strength of the smile), and smile confidence for each detected face.

Accepted Answer

Since the tool is not perfect, additional performance gains could be obtained by incorporating other modalities, such as the overhead Kinect view.

Accepted Answer

In addition, for each stage of the protocol, she rates the effort required to engage the child using a 3-point Likert scale, with a score of 0 indicating that the child was easily engaged and a score of 2 indicating that significant effort was required.

Accepted Answer

The authors employed a late-fusion approach to combining modalities, and worked directly with the previously-trained audio and video classifiers.

Accepted Answer

Using 8 sessions with high parser confidence, the authors correctly detected smiles in 31 out of 40 substages, giving a balanced accuracy of 76.6%.

Accepted Answer

In their experiments, the groups of features retained by the decision tree for the Ball stage were {Visibility, Near Examiner}, {Shown, Touched Examiner} and {Touched Child}.

Accepted Answer

The authors then computed a rotation-invariant region covariance descriptor for each template, and identified the two that were the most discriminative.

Decoding Children's Social Behavior

Chat with Paper

AI Agents for this Paper

Most frequently asked questions

1. What are the contributions mentioned in the paper "Decoding children’s social behavior" ?

2. How can the authors obtain information about the state of the dyad?

3. What is the contribution of this paper?

4. What is the main source of error in the tracker?

5. How many children have participated in the Rapid-ABC assessment?

6. How many bins did the authors select for the 2D histogram and pyramid level?

7. How many false positives did the Nexidia tool detect in the first 30 seconds of the stage?

8. How many sessions did the examiner stay at the table during the analysis?

9. What is the way to detect a smile?

10. What is the way to improve the performance of the tool?

11. How does the examiner rate the effort required to engage the child?

12. What was the method used to combine the audio and video features?

13. How did the authors detect smiles in the substages?

14. What were the groups of features retained by the decision tree for the Ball stage?

15. What was the discriminative descriptor for each template?

Figures

Citations

3D skeleton-based human action classification

Panoptic Studio: A Massively Multiview System for Social Interaction Capture

Applying Machine Learning to Facilitate Autism Diagnostics: Pitfalls and Promises

Emotion recognition in the wild from videos using images

Extending Long Short-Term Memory for Multi-View Structured Learning

References

Learning realistic human actions from movies

Opensmile: the munich versatile and fast open-source audio feature extractor

Actions as space-time shapes

Actions as Space-Time Shapes

Region covariance: a fast descriptor for detection and classification

Related Papers (5)

The Extended Cohn-Kanade Dataset (CK+): A complete dataset for action unit and emotion-specified expression

On Space-Time Interest Points

A Survey of Affect Recognition Methods: Audio, Visual, and Spontaneous Expressions

Robust Real-Time Face Detection

The autism diagnostic observation schedule-generic: a standard measure of social and communication deficits associated with the spectrum of autism.