Conversation
lucaslibshutz
requested changes
May 18, 2026
lucaslibshutz
left a comment
lucaslibshutz
left a comment
Contributor
There was a problem hiding this comment.
good work overall!! just need to fix some minor formatting things and maybe add some formulas (up to you though, if its too complicated just add one or two, or maybe some numerical results to quantify how you found Canny edge detection was worse).
| \end{itemize} | ||
|
|
||
| \subsubsection*{Lane Segmentation Architecture} | ||
| \begin{itemize} |
Contributor
There was a problem hiding this comment.
these should all be changed to \paragraph so that they don't overflow, see what @ThomasXin50 did for his version.
|
|
||
| \item \textbf{Traditional CV vs. Deep Learning:} Initially, we attempted a classical computer vision approach using Canny Edge Detection coupled with a Hough Transform. While this method was computationally inexpensive, it proved extremely brittle in unpredictable outdoor environments. Shadows, sun glare, and degraded paint easily broke the edge detection thresholds, and the Hough Transform frequently misidentified the edges of cones, barrels, or grass boundaries as lane lines. We pivoted to a Machine Learning approach because convolutional neural networks learn semantic context, allowing them to accurately segment drivable paths and lanes even when the physical lines are partially obscured or subjected to harsh lighting gradients. | ||
|
|
||
| \item \textbf{Model Deep Dive (\texttt{LFD\_Roadseg}):} For our core lane detection, we utilized the \texttt{LFD\_Roadseg} architecture, which employs a dual-stream design to balance high-speed inference with accurate spatial resolution. As shown in Figure \ref{fig:lfd_model}, the model processes two inputs simultaneously: a high-resolution image ($H \times W$) to preserve fine, pixel-level details like thin lane boundaries, and an aggressively downsampled low-resolution image to quickly capture the broader, global context of the road. |
Contributor
There was a problem hiding this comment.
this is sweet, some formulas would be nice to help quantify what you're talking about w.r.t the image detection stuff.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
No description provided.