A Hybrid Grammar-Based Approach for Learning and Recognizing Natural Hand Gestures
Authors: Amir Sadeghipour, Stefan Kopp
AAAI 2014 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We applied the FSCFG framework to learn a generalized model from the gesture performances performed for each of the twenty objects in the 3DIG dataset. The learned grammar models were then tested by classifying the unseen gesture performances, based on their parsing probabilities. Figure 5 shows the receiver operating characteristics (ROC) graphs of two-fold cross-validation results, for each subset of the gesture performances. As shown in Figure 7, the FSCFG model outperforms the other algorithms in all subsets of the gesture dataset. |
| Researcher Affiliation | Academia | Amir Sadeghipour and Stefan Kopp Faculty of Technology, Center of Excellence Cognitive Interaction Technology (CITEC), Bielefeld University, P.O. Box 100131, D-33501 Bielefeld, Germany |
| Pseudocode | No | The paper describes the algorithms and processes verbally and through mathematical equations but does not include any clearly labeled pseudocode or algorithm blocks. |
| Open Source Code | No | The paper refers to third-party libraries used (Bayes Net Toolbox for Matlab, LIBSVM library) and provides links to them, but it does not state that the authors' own code for the proposed FSCFG framework is open-source or provide a link to it. |
| Open Datasets | Yes | This three-dimensional iconic gesture dataset (3DIG) has been made available online1. 1http://projects.ict.usc.edu/3dig/ |
| Dataset Splits | Yes | Figure 5 shows the receiver operating characteristics (ROC) graphs of two-fold cross-validation results, for each subset of the gesture performances. |
| Hardware Specification | No | The paper mentions using "MS Kinect TM" for recording the dataset but does not specify any hardware (CPU, GPU, memory) used for running the computational experiments or training the models. |
| Software Dependencies | No | The paper mentions using "the Bayes Net Toolbox for Matlab" and "the LIBSVM library (Chang and Lin 2011)". While it cites the LIBSVM publication, it does not provide specific version numbers for these software packages, nor for Matlab itself. |
| Experiment Setup | Yes | In this application of FSCFG, applying the na ıve but fast best-first search algorithm with two look-ahead steps, we could achieve classification results as accurate as the multilevel best-first and beam search algorithms. |