The embodiment of the invention provides a target area detection method and device, electronic equipment and a readable storage medium, and the method comprises the steps: inputting a target picture into a convolution layer, obtaining a first prediction vector, and enabling the convolution layer to be formed by the connection of one or more convolution kernels of a low semantic layer and one or more convolution kernels of a high semantic layer; inputting a first prediction vector output by the convolution kernel of the low semantic layer into a context sensitive prediction layer to obtain a second prediction vector, the context sensitive prediction layer comprising a cascade structure of two or more preset convolution kernels, and the number of variables of the preset convolution kernels in each dimension being greater than 1; inputting the first pre-vectors output by the convolution kernels of the high semantic layer into a representative feature extraction layer to obtain third prediction vectors, wherein the representative feature extraction layer only comprises one preset convolution kernel; and determining a target area according to the second prediction vector and the third prediction vector. The complexity of a network structure can be effectively reduced in a high semantic layer, and an over-fitting phenomenon is avoided.