Individual pig object detection algorithm based on Gaussian mixture model

Abstract

The background models are crucially important for the object extraction for moving objects detection in a video. The Gaussian mixture model (GMM) is one of popular methods in the background models. Gaussian mixture model which applied to the pig target detection has some shortcomings such as low efficiency of algorithm, misjudgment points and ghosts. This study proposed an improved algorithm based on adaptive Gaussian mixture model, to overcome the deficiencies of the traditional Gaussian mixture model in pig object detection. Based on Gaussian mixture background model, this paper introduced two new parameters of video frames m and T 0. The Gaussian distribution was scanned once every m frames, the excessive Gaussian distribution was deleted to improve the convergence speed of the model. Meanwhile, using different learning rates to suppress ghosts, a higher decreasing learning rate was adopted to accelerate the background modeling before T 0 , the background model would become stable as the time continued and a smaller learning rate could be used. In order to maintain a stable background and reduce noise interference, a fixed learning rate after T 0 was used. Results of experiments indicated that this algorithm could quickly build the initial background model, detect the moving target pigs, and extract the complete contours of the target pigs'. The algorithm is characterized by good robustness and adaptability.

The Application of Cameras in Precision Pig Farming: An Overview for Swine-Keeping Professionals

“…The identification process has been performed with object detection from an image, then identifying the individual pig from the detected objects using morphological characteristics. Various studies used specific techniques such as GMM-based background subtraction [20,68,69], denoising using low-pass filtering followed by Otsu's threshold method, morphological operations, ellipse fitting [21,28,64], graphical module-based segmentation [70][71][72], and learning-based tracking [65] for identification.…”

“…Similarly, motion-based video monitoring for ASF detection was conducted in a previous study [112]. A typical 2D camera was used to record the videos; the experiment had four stages: ASF free (days 1-11), infection time (days 12-15), qPCR detection (days [16][17][18], and clinical detection (days [19][20][21][22][23]. All the stages were completed.…”

“…Information and individual monitoring-maximum data could be stored in a computer, which could be used to further visualize the situation (feeding efficiency can be calculated by collecting data of feed intake and the weight of the pigs [17][18][19]; body surface temperature data can be used to track heat stress [15]). All the data, including on physical movement, performance, biological information, phonotype properties, etc., can be used to visualize not only group-or house-level data but individual data, too [20][21][22].…”

“…Analyzing ability-with big data, it is possible to produce advanced-level software to notify farmers about standard criteria [20][21][22].…”