Abnormal driving behavior identification method based on CNN-LSTM spatial-temporal feature fusion

The invention discloses an abnormal driving behavior recognition method based on CNN-LSTM spatial-temporal feature fusion, and the method comprises an image collection processing system, a convolutional neural network layer, a long-short-term memory network layer, a full connection layer and a Softmax layer, and also comprises the steps: 1, preliminary feature image extraction; 2, grouping of thepreliminary feature images; step 3, spatial feature extraction; 4, time sequence feature extraction and fusion; step 5, identification and judgment; according to the method, spatial features and timesequence features of multiple frames of driver abnormal behavior images are extracted through the convolutional neural network layer and the long-short-term memory network layer, and fusion transmission is carried out; the driving behavior state of the driver is more accurately identified, the full connection layer and the Softmax layer are utilized to identify and determine the abnormal behaviorimage of the driver, compar