OpenCV: Inconsistent Results with Optical Flow Calculation Using Farneback Method
I've been struggling with this for a few days now and could really use some help. I'm updating my dependencies and I've searched everywhere and can't find a clear answer. I'm working on a project where I need to analyze the motion of objects in a video stream using OpenCV's Farneback method for optical flow calculation. However, I'm observing inconsistent results when processing videos with different frame rates and resolutions. For example, when I run the following code on a 30 FPS video with a resolution of 640x480, I get smooth optical flow vectors, but when I switch to a 60 FPS video with a resolution of 1280x720, the flow vectors appear jagged and sometimes even point in the opposite direction of the actual motion. Here’s the code I’m using: ```python import cv2 import numpy as np cap = cv2.VideoCapture('input_video.mp4') # Parameters for Farneback optical flow farneback_params = { 'pyr_scale': 0.5, 'levels': 3, 'iterations': 3, 'poly_n': 5, 'poly_sigma': 1.2, 'flags': 0 } ret, old_frame = cap.read() old_gray = cv2.cvtColor(old_frame, cv2.COLOR_BGR2GRAY) while True: ret, frame = cap.read() if not ret: break gray = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY) flow = cv2.calcOpticalFlowFarneback(old_gray, gray, None, **farneback_params) # Visualizing the flow hsv = np.zeros_like(frame) magnitude, angle = cv2.cartToPolar(flow[..., 0], flow[..., 1]) hsv[..., 0] = angle * 180 / np.pi / 2 hsv[..., 2] = cv2.normalize(magnitude, None, 0, 255, cv2.NORM_MINMAX) rgb = cv2.cvtColor(hsv, cv2.COLOR_HSV2BGR) cv2.imshow('Optical Flow', rgb) old_gray = gray.copy() if cv2.waitKey(30) & 0xFF == 27: break cap.release() cv2.destroyAllWindows() ``` I’ve tried adjusting the `pyr_scale` and `poly_n` parameters to see if I could get more stable results, but it doesn't seem to help much. I also checked that the videos are properly encoded and not corrupted. I'm running OpenCV version 4.5.1 on Python 3.8, and I’m curious if there are any specific settings or best practices for using optical flow with videos at varying frame rates and resolutions. Is it possible that the parameters I’m using are not suitable for higher resolution videos? Any suggestions would be greatly appreciated! Any ideas what could be causing this? I'm working on a microservice that needs to handle this. Any help would be greatly appreciated! Any suggestions would be helpful.