The image is preprocessed by multi-structure closed operation reconstruction. A small target region extraction algorithm combining Selective Search algorithm and Harris corner detection algorithm is proposed

which eliminates a lot of false targets

reduces the number of candidate frames

and improves the speed

the non-maximum suppression（ NMS） algorithm is used to remove the overlapping small target candidate frame

and the deep forest algorithm（ gc Forest） is improved

the depth structure is added to the multi-granularity scanning part

and more information is transmitted to the cascade forest part. Even in small target areas where data sets are scarce

good results can be obtained. The improved deep forest algorithm is used to detect small targets on different datasets.Experiments show that the method improves the accuracy and speed of detecting small targets.