A tumor is one of the major concerns of the modern-day society, affecting a large group of the population. These are difficult to cure at later stages but if detected earlier can be cured in many cases. Out of these vast classes of tumors is a class Soft Tissue Tumor (STTs). STTs are a heterogeneous group of neoplasms that can arise from any of the soft tissues of the body. They are relatively rare but can be aggressive and life-threatening. Accurate diagnosis of STTs is essential for appropriate treatment planning. Traditionally, diagnosis of STTs is based on a combination of clinical findings, imaging studies, and biopsy. However, these methods can be challenging, especially for STTs that are small, poorly differentiated, or located in difficult-to-access areas. One of the major parts of curing a tumor patient is its detection. This paper presents different machine learning (ML) based approach tailored for the detection of STTs. Although tumor detection has projected many challenges because of their small sizes (at initial stages) and similar appearances as of the surrounding tissues. Here we described and compared the various techniques of identification.