Video Encoder with Target Detection Function Based on Yolov5

Zhibin Liu

《World Scientific Research Journal》 8卷4期 (2022/04) Pp. 302-308

https://doi.org/10.6911/WSRJ.202204_8(4).0039

摘要

Ultra low latency transmission, high bandwidth and ubiquitous access points have greatly alleviated the ＂last mile＂ problem of streaming media live service. However, the complex compression coding of high‐resolution video is still inevitable before transmission. We combine the YOLOv5 target detection algorithm with the video encoder, segment the foreground and background of the video to be encoded with the help of the improved YOLOv5, and then encode the foreground and background respectively. The foreground area of the video uses the normal motion search to find the motion vector, and the background area uses the global motion parameters to describe its overall motion. This scheme can get the target detection results and encode the video at the same time. The experimental results show that the proposed scheme greatly reduces the complexity of the encoder with less loss of video quality, up to 82%. At the same time, the target detection task is completed.

關鍵字

Video coding ； Target detection ； Low delay ； Joint coding

參考文獻

R. Khemiri, N. Bahri, F. Belghith, et al., “Fast motion estimation for hevc video coding,” in Image Processing, Applications & Systems, 2017.

W. Chien, K. Liao, and J. Yang, “Enhanced amvp mechanism based adaptive motion search range decision algorithm for fast hevc coding,” in 2014 IEEE International Conference on Image Processing (ICIP), pp. 3696–3699, 2014.

H. Chen, J. Ding, and Y. Lee, “Novel adaptive rood path searches with small motion prejudgments for fast block motion estimation,” in 2018 IEEE Asia Pacific Conference on Circuits and Systems (APCCAS), pp. 560–563, 2018.

X. Li, R. Wang, X. Cui, et al., “Context-adaptive fast motion estimation of hevc,” in 2015 IEEE International Symposium on Circuits and Systems (ISCAS), pp. 2784–2787, 2015.

S. Hsia and Y. Hung, “Fast multi-frame motion estimation for h264/avc system”, Signal Image & Video Processing, 4(2):167–175,2010.