Blockchain, Machine Learning, and the Internet of Things are three of the areas where research is currently being undertaken. Suspicious object detection techniques are discussed as well. Researchers have resorted to machine learning to develop a highly trained network that can identify situations that are analogous. Before the machine learning approach is used to analyze the image and find the pattern, each frame is gathered and the data is preprocessed to reduce the size of the picture. Here, we use the pertained set to keep tabs on a live activity. The study article included previous studies that focused on the same problem. We have spent a lot of time discussing the issues with the earlier research. We also investigate the many recommended therapies and methods of solving the problem. In order to narrow down the search for a malicious IoT component, our research used deep learning and image processing. The simulation findings reveal that PSNR, MSE, and SSIM values change depending on the noise-reduction technique used. The Gaussian filter yields the highest PSNR, whereas the bilateral filter yields the lowest. While the Gaussian filter has the lowest MSE, the blur filter has the highest. A Gaussian filter is used to increase the output of an SSIM simulation, whereas a blur filter is used to limit it. After detection of suspicious activity records are managed in blockchain for integrity management. The present research is capable of finding suspicious objects from running traffic and storing traceability by making integration of blockchain.