Using syllabic Mel cepstrum features and k-nearest neighbors to identify anurans and birds species

Vaca-Castano, Gonzalo; Rodriguez, Domingo

doi:10.1109/sips.2010.5624892

Cited by 20 publications

(27 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Therefore, a small amount of noise can cause erroneous classification by SVM. This also explains why MFCC usually needs other carefully selected features [37] to improve classification performance. Our`1 classifier performs much better than the other two classifiers, especially when SNR is low.…”

Section: Comparing To Benchmarksmentioning

confidence: 99%

“…The testbed, which is located on our campus with thin vegetation, consists of five nodes (Figure 8(a)) configured as Ad-hoc mode with a star network topology. The aim of the experiments is automatic bird vocalization recognition, which is a typical pattern recognition problem [32,13,37]. We choose two bird species that are frequently observed on our campus: cockatoos (Figure 8(b)) and rainbow lorikeets (Figure 8(c)).…”

Section: Experiments On Testbedmentioning

confidence: 99%

“…However, the selection of good features is not a trivial problem and the performance of the classification algorithm often depends on the selection of good features. Moreover, di↵erent feature selections can result in significantly di↵erent classification accuracy for certain acoustic signals, ranging from 50% to 91.53% [34,24,37]. Although Mel-frequency Cepstrum Coe cients (MFCC) is one of the most popular features which, together with SVMs, has been used successfully in ASNs [13,24,32], MFCC usually needs other carefully selected features [37] to improve classification performance.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Real-time classification via sparse representation in acoustic sensor networks

Wei

Yang

Shen

et al. 2013

Proceedings of the 11th ACM Conference on Embedded Networked Sensor Systems

View full text Add to dashboard Cite

Acoustic Sensor Networks (ASNs) have a wide range of applications in natural and urban environment monitoring, as well as indoor activity monitoring. In-network classification is critically important in ASNs because wireless transmission costs several orders of magnitude more energy than computation. The main challenges of in-network classification in ASNs include e↵ective feature selection, intensive computation requirement and high noise levels. To address these challenges, we propose a sparse representation based featureless, low computational cost, and noise resilient framework for in-network classification in ASNs. The key component of Sparse Approximation based Classification (SAC),`1 minimization, is a convex optimization problem, and is known to be computationally expensive. Furthermore, SAC algorithms assumes that the test samples are a linear combination of a few training samples in the training sets. For acoustic applications, this results in a very large training dictionary, making the computation infeasible to be performed on resource constrained ASN platforms. Therefore, we propose several techniques to reduce the size of the problem, so as to fit SAC for in-network classification in ASNs. Our extensive evaluation using two real-life datasets (consisting of calls from 14 frog species and 20 cricket species respectively) shows that the proposed SAC framework outperforms conventional approaches such as Support Vector Machines (SVMs) and k-Nearest Neighbor (kNN) in terms of classification accuracy and robustness. Moreover, our SAC approach can deal with multi-label classification which is common in ASNs. Finally, we explore the system design spaces and demonstrate the real-time feasibility of the proposed framework by the implementation and evaluation of an acoustic classification application on an embedded ASN testbed.

show abstract

Section: Comparing To Benchmarksmentioning

confidence: 99%

Section: Experiments On Testbedmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Real-time classification via sparse representation in acoustic sensor networks

Wei

Yang

Shen

et al. 2013

Proceedings of the 11th ACM Conference on Embedded Networked Sensor Systems

View full text Add to dashboard Cite

show abstract

“…The classification results using automatic techniques range from 50% in [19], 82.6% in [18] and 99.3% in [9]. The large difference between the results is mainly due to the information used as base for the classification.…”

Section: B Anuran Classificationmentioning

confidence: 99%

“…Previous efforts for automatic animal classification, based on audio streams, have pointed out some features and classifiers for a successful task [9]. The list of features includes Spectral Centroid (S), Signal Bandwidth (B), Zero-crossing Rate (ZC), Hybrid Spectral-Entropy (H) and Mel Fourier Cepstral Coefficient (MFCCs).…”

Section: Introductionmentioning

confidence: 99%

On the Use of Compressive Sensing for the Reconstruction of Anuran Sounds in a Wireless Sensor Network

Diaz

Nakamura²,

Yehia

et al. 2012

2012 IEEE International Conference on Green Computing and Communications

View full text Add to dashboard Cite

Natural sounds like wind, water, wildlife, and vegetation are considered acoustic sounds and are referred to as the soundscape. Wildlife sounds usually provide enough data to classify and monitor the fauna. In this context, anurans (frogs and toads) have been used by biologists as early indicators of ecological stress in a given environment. Compressive sensing is a promising technique that can be used to reduce the amount of transmitted data in a wireless sensor network, and, thus, minimize the limited resources of sensors nodes. In this work, we collect samples of the anuran audio using compressive sensing, send them towards the sink node, where the samples are reconstructed and identify the correct anuran specie of each call. Our main goal is to evaluate whether compressive sensing is a viable technique to reconstruct properly the anuran calls and identify the correct specie in an environment monitored by a wireless sensor network. We evaluate the proposed methodology by calculating different sound distances between the original and the reconstructed calls. Results show that sampling only 10% of the original data, the sink node can reconstruct the original audio with a high quality.

show abstract

How to Correctly Evaluate an Automatic Bioacoustics Classification Method

Colonna

Gama

Nakamura

2016

Advances in Artificial Intelligence

View full text Add to dashboard Cite

In this work, we introduce a more appropriate (or alternative) approach to evaluate the performance and the generalization capabilities of a framework for automatic anuran call recognition. We show that, by using the common k-folds Cross-Validation (k-CV) procedure to evaluate the expected error in a syllable-based recognition system the recognition accuracy is overestimated. To overcome this problem, and to provide a fair evaluation, we propose a new CV procedure in which the specimen information is considered during the split step of the k-CV. Therefore, we performed a k-CV by specimens (or individuals) showing that the accuracy of the system decrease considerably. By introducing the specimen information, we are able to answer a more fundamental question: Given a set of syllables that belongs to a specific group of individuals, can we recognize new specimens of the same species? In this article, we go deeper into the reviews and the experimental evaluations to answer this question.

show abstract

Using syllabic Mel cepstrum features and k-nearest neighbors to identify anurans and birds species

Cited by 20 publications

References 13 publications

Real-time classification via sparse representation in acoustic sensor networks

Real-time classification via sparse representation in acoustic sensor networks

On the Use of Compressive Sensing for the Reconstruction of Anuran Sounds in a Wireless Sensor Network

How to Correctly Evaluate an Automatic Bioacoustics Classification Method

Contact Info

Product

Resources

About