Repository navigation
Conversation
The face landmarker is constructed with RunningMode::VIDEO. With the bundled mars-face-kit, that mode never emits landmarks for single-frame Detect() calls (a video tracker needs a run of frames to converge, and FaceDetector::Detect passes each frame with timestamp 0), so every landmark-driven filter — face reshape, big-eye, lipstick, blusher — silently no-ops. Switching to RunningMode::IMAGE returns 111 landmarks on the first call. Also surface the mars Init() return code: it was discarded, and LOG_* writes to std::cout which is /dev/null on Android, so a model that failed to load was indistinguishable from a working detector that finds no faces.
This branch has not been deployed
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Problem
The face landmarker is constructed with
RunningMode::VIDEO. With the bundled mars-face-kit, that mode never emits landmarks for single-frameDetect()calls — a video tracker needs a run of frames to converge, andFaceDetector::Detectpasses each frame with timestamp 0. As a result every landmark-driven filter (face reshape, big-eye, lipstick, blusher) silently produces no effect.Fix
Construct the landmarker with
RunningMode::IMAGE, which returns 111 landmarks on the first call. Also surface the marsInit()return code: it was discarded, andLOG_*writes tostd::cout(which is/dev/nullon Android), so a model that genuinely failed to load was indistinguishable from a working detector that finds no faces.Notes
Verified on Android (Snapdragon 845): landmarks return on the first call and the reshape/makeup filters take effect.