Child Sexual Abuse Material (CSAM) is defined as audiovisual content depicting nudity and/or sexual activity involving a minor. The development of an effective application based on machine learning for the identification of CSAM is critical in the field of forensic sexology. Such a solution, however, must adhere to the European Commission’s guidelines, including the requirement to present explainable classification results that can be understood by experts. To address this need, a study was conducted to develop machine learning models capable of detecting objects, such as anatomical structures, that are pertinent to CSAM identification. For this study, a forensic expert trained in anthropology and sexology classified 5,000 pornographic images featuring both minors and adults. The study was conducted with the approval of the ethics committee. Seven experiments were performed using Apple’s CoreML framework, testing fundamental hyperparameters related to model training. This research represents a significant step towards creating ML applications that not only detect CSAM effectively but also provide explainable information regarding the predictions made, aligning with regulatory and expert requirements.