Marine Pollution has emerged as a serious challenge for the world which endangers coastal ecosystem, marine life, and people’s livelihoods. This research utilizes Sentinel-2, Machine Learning (ML) and Google Earth Engine (GEE) to monitor and quantify the presence of floating plastic debris along the coastline of Odisha, with focus on Mahanadi River, Podampetta, South Jamboo (Kendrapara), Satapara and Rambha in Chilika Lake. These plastic pollution hotspots are well known for their ecological sensitivity and receive significant plastic pollution from riverine discharge and human activities. This study classified floating plastics using Random Forest (RF) ML models in GEE based on the Floating Debris Index (FDI) and other spectral indices (NDVI, PI, kNDVI), together with targeted spectral bands (Red Edge 2, NIR, SWIR1).Post-Monsoon observation at Podampetta revealed high concentration of plastic debris. Pre-Monsoon examination at Satapara foundthat chilika lake had accumulated plastic waste, attributed to residual river discharge and tourism. Models 4 and 5 successfully identified near-shore and scattered patterns throughout seasons, demonstrating the seasonal fluctuations in plastic trash in Rambha. Due to the influence of urban and industrial garbage, the Mahanadi River showed significant plastic loads, especially during the monsoon season. Model 1,2, and 3 successfully mapped plastic debris, while Model4 and 5 demonstrated superior debris detection during peak seasons. South Jamboo (Kendrapara) received plastic primarily via riverine inputs.Models 1, 2 and 3 consistently mapped shoreline plastic patches. The RF model consistently mapped floating plastic across diverse environments. While Models 1–3 worked efficiently in highly polluted and seasonally variable locations (e.g., Mahanadi River, Podampetta), Models 4 and 5 proved more effective in detecting dispersed debris in open-water environments like Chilika Lake. Field validation of the model was conducted and was found correct. The RF configurations were very accurate in detection of macroplastics as well as fragmented plastic waste. GEE exceeded expectations in providing tools for large scale analysis, proving its efficiency and effectiveness in monitoring rapidly changing coastal ecosystems. This study shows the effectiveness of combining remote sensing, spectral indices, and machine learning for marine plastic pollution. It allows the detection of zones where actions need to be focused, taking into account the hot spots and zones of seasonal accumulation, enabling the policymakers as well as environmental organizations to take appropriate measures to protect sensitive coastal regions.