Instance segmentation of objects with lane detection on road.
<p>Instance segmentation is extensively found in applications of medical imaging, object detection and many computer vision applications.</p>
<p>One such application is this project where vehicles or different object surrounding the car are detected and identified which belong to a particular class. This project was implemented on the <a href="https://medium.com/@jaimin-k/detecting-lanes-on-road-46700c32ebe5" rel="noopener"><em>Lane detection</em></a> project which I had completed earlier.</p>
<p><img alt="" src="https://miro.medium.com/v2/resize:fit:630/1*Kw1e_pKVL6D5u86i0Fa53g.png" style="height:379px; width:700px" /></p>
<p>A frame from the output video</p>
<p>Due to computational limitations, there was a lag in the output video and was taking much more time to process. Above is a frame from the processed video.</p>
<p>The project uses <strong>PixelLib</strong> library for instance segmentation, which is based on the MaskRCNN framework.</p>
<p><a href="https://medium.com/@jaimin-k/instance-segmentation-of-objects-with-lane-detection-on-road-dcc9f5514de0"><strong>Website</strong></a></p>