I need an image segmentation solution that can automatically isolate industrial parts appearing in a video stream. The footage will be delivered as standard video files (e.g., MP4), so your workflow should include extracting frames and then applying the segmentation model. What I’m after • Pixel-accurate segmentation masks for each industrial part in every frame. • A Python script (preferably leveraging OpenCV, TensorFlow or PyTorch) that loads the video, processes it frame-by-frame, and outputs the masks—either as PNG overlays or a reconstructed video with alpha channels. • Clear instructions on how to run the code on a typical Windows or Linux workstation. Nice-to-have extras • Basic post-processing to remove noise and smooth mask edges. • A short README explaining any model-training steps, hyper-parameters, and how to adapt the code to new parts or camera angles. Please keep the solution self-contained, relying only on publicly available libraries. Once delivered, I’ll test it on a fresh set of videos; if the masks align correctly with the parts, the job is complete.