The ubiquity of digital manipulation tools in the public domain exposes several domains like forensics, journalism, and arts to a great deal of risks because image forgery has become a common feature. While there is high demand for the detection of forged images, machine learning turns in excellent performance based on signs so subtle that indicate tampering. In this paper, some of the most recent advances in detecting image forgery will be presented, where detection relies on different types of forgeries like copy-move, splicing, and retouching. We review traditional approaches used, such as SVM, CNN, and Random Forests, along with the assessment of SIFT and CNN-based feature extraction methodologies. Further, this includes insights into the important data sources used for the training and testing of models, identifying major lacunae in the methodologies presently undertaken. The review presents, amidst emerging challenges like deepfake forging and others, the need to look out for real-time, robust detection methodologies with more functionalities and capabilities, with the hope of aiding the digital image forensics research community by reinforcing the trust of mankind in digital images amidst increasing propaganda with misinformation.