Snake Detection and Blurring System to Prevent Unexpected Appearance of Snake Images on Visual Media Sources Using Deep Learning

Abstract

This paper proposes an object detection and blurring system using deep learning especially for the detection of snake images since while some of the people does not care about the appearance of the snake images on the visual sources, there is a significant portion of them who are uncomfortable of seeing snakes as well as their images. Therefore, with the system that is proposed in this paper, the snake objects from the images would be detected with YOLOv4 trained model and blurred using OpenCV. The detection f1 score of the selected model is 92% on the test set.

Keywords

• blur
• deep learning
• object detection
• phobia
• snake

References

1. [1] E. Landova, J. Maresova, O. Simkova, V. Cikanova and D. Frynta, "Human responses to live snakes and their photographs: Evaluation of beauty and fear of the king snakes," Journal of Environmental Psychology, vol. 32, no. 1, pp. 69-77, 3 2012.
2. [2] A. Bochkovskiy, C.-Y. Wang and H.-Y. M. Liao, "YOLOv4: Optimal Speed and Accuracy of Object Detection," 2020.
3. [3] M. Tan, R. Pang and Q. V. Le, "EfficientDet: Scalable and Efficient Object Detection," 2020.
4. [4] OpenCV team, "About," OpenCV, 2021. [Online]. Available: https://opencv.org/about/. [Accessed 20 May 2021].
5. [5] Google Inc., "Open Images V6 Snake Category Detection Type," 2021. [Online]. Available: https://storage.googleapis.com/openimages/web/visualizer/index.html?set=train&type=detection&c=%2Fm%2F078jl. [Accessed 20 May 2021].
6. [6] Google Inc., "Overview of Open Images V6," Google Inc., 2021. [Online]. Available: https://storage.googleapis.com/openimages/web/factsfigures.html. [Accessed 21 May 2021].
7. [7] AlexeyAB, "Yolo v4, v3 and v2 for Windows and Linux," 15 5 2021. [Online]. Available: https://github.com/AlexeyAB/darknet. [Accessed 21 May 2021].