google-ai-edge/mediapipe issues and pull requests

#5638 - TypeError: Super constructor null of nJ is not a constructor (for ImageSegmenter initialization)

Issue - State: open - Opened by vastava about 13 hours ago

#5637 - Added audio file utils and options tests for iOS Audio Classifier

Pull Request - State: closed - Opened by priankakariatyml about 16 hours ago

#5636 - Fix for isIOS() platform util on worker and non-worker contexts

Pull Request - State: open - Opened by joacopaz 1 day ago

#5635 - Fixed leading white space stripping and added new options to iOS LlmInference

Pull Request - State: open - Opened by priankakariatyml 2 days ago

#5634 - Extract blendshape or landmark using flame-based mesh as input

Issue - State: open - Opened by a1e0-heewon 2 days ago
Labels: type:feature

#5633 - how to build com.google.mediapipe:tasks-genai for android..need steps please

Issue - State: open - Opened by BalajiPolisetty2207 2 days ago - 8 comments
Labels: type:build/install

#5632 - Language Dedtection documents link is broken

Issue - State: open - Opened by sajjadrahman56 2 days ago - 1 comment
Labels: type:docs-bug, platform:javascript, task:language detection

#5631 - Running tasks-vision on Mac Chrome Web Worker crashes

Issue - State: open - Opened by joacopaz 3 days ago - 5 comments
Labels: platform:javascript, type:bug, stat:awaiting googler

#5630 - Conversion of TFLite model precision and original model files.

Issue - State: open - Opened by Criminal-9527 3 days ago - 1 comment
Labels: type:support, type:tflite, task:hand landmarker

#5629 - Update OpenCV version

Pull Request - State: open - Opened by j20001970 4 days ago

#5628 - Update the docs to show how to build tasks_vision Android

Issue - State: closed - Opened by calbot 6 days ago - 9 comments
Labels: type:build/install, platform:android, os:linux-non-arm

#5627 - Docker mediapipe image build fails

Issue - State: closed - Opened by calbot 6 days ago - 3 comments
Labels: type:build/install

#5626 - extensive memory usage while running mediapipe hand landmarker and custom object detection

Issue - State: open - Opened by AlundorZhu 7 days ago - 2 comments
Labels: task::all, type:performance, task:hand landmarker, os:macOS

#5625 - Cross-compile mediapipe for ARM64 on SemiDrive X9U platform

Issue - State: open - Opened by Sirhxs 7 days ago - 2 comments
Labels: type:support, type:build/install, stat:awaiting response, platform:embedded-linux-arm

#5624 - Android/Java PoseLandmarker unnecessary []byte garbage

Issue - State: open - Opened by calbot 7 days ago - 2 comments
Labels: platform:android, type:performance, task:pose landmarker

#5623 - Added iOS Audio Embedder header and implementation

Pull Request - State: closed - Opened by priankakariatyml 8 days ago

#5622 - Explanation of output shape [1, 117] (World landmarks for pose) or [1, 195] (Pose landmarks) of pose_landmarks_detector.tflite in Mediapipe

Issue - State: open - Opened by mbkamran 9 days ago
Labels: type:support, platform:python, task:pose landmarker

#5621 - Added few basic audio clip tests for iOS Audio Classifier

Pull Request - State: closed - Opened by priankakariatyml 10 days ago

#5620 - Help!!! How to save the current training checkpoint in colab

Issue - State: open - Opened by ZFbaby 10 days ago - 2 comments
Labels: stat:awaiting response, platform:python, stale, type:modelmaker, task:object detection, os:macOS

#5619 - @mediapipe/tasks-vision pollutes browser console with logs

Issue - State: closed - Opened by mikoprus 11 days ago - 3 comments
Labels: platform:javascript, type:bug, task:face detection, os:macOS

#5618 - Mediapipe don't work with 'VIDEO' mode on Android WebView while 'IMAGE' mode work very well.

Issue - State: closed - Opened by ffpwserver 11 days ago - 2 comments
Labels: platform:android, platform:javascript, type:bug, task:gesture recognition, task:face landmarker

#5617 - Model Maker Supportting for Image Segmentation

Issue - State: closed - Opened by consistent1997 12 days ago - 4 comments
Labels: type:feature, platform:python, type:modelmaker, task:image segmentation, os:linux-non-arm

#5616 - What is the output of the `hand_detector.tflite` file using the hand landmark detection model `hand_landmark.task` locally?

Issue - State: open - Opened by Criminal-9527 12 days ago
Labels: type:support, type:tflite, platform:python, task:hand landmarker

#5615 - Can't find hand_landmark.tflite

Issue - State: closed - Opened by Criminal-9527 14 days ago - 1 comment
Labels: type:others

#5614 - object detection on android with gpu will crash when set setCategoryAllowlist

Issue - State: open - Opened by hugski 15 days ago - 3 comments
Labels: platform:android, gpu, type:bug, task:object detection

#5613 - Lack of documentation for compiling MediaPipe as a TensorFlow.js WASM package

Issue - State: open - Opened by ginesmartinezros 15 days ago
Labels: type:support, legacy:face mesh, platform:javascript, type:docs-feature

#5612 - Why mediapipe v0.10.15 depends on numpy (<2) and v0.10.14 doesn't?

Issue - State: open - Opened by alanwilter 15 days ago - 3 comments
Labels: type:support, task::all, stat:awaiting response, platform:python, dependencies

#5611 - I can modify the recognized categories by modifying the calculator（TfLiteTensorsToDetectionsCalculator）

Issue - State: open - Opened by tisaud 16 days ago
Labels: type:others

#5610 - Support Gemma2-2b model for inference in Android

Issue - State: open - Opened by FranzKafkaYu 16 days ago - 10 comments
Labels: type:feature, platform:android, task:LLM inference

#5609 - INTERNAL: Service "kGpuService", required by node mediapipe_tasks_vision_pose_landmarker_poselandmarkergraph__mediapipe_tasks_vision_pose_detector_posedetectorgraph...

Issue - State: open - Opened by V-m1r 16 days ago - 1 comment
Labels: type:support, gpu, platform:javascript, task:pose landmarker

#5608 - Running on a TV box with poor performance

Issue - State: closed - Opened by InitialFStudio 16 days ago - 4 comments
Labels: platform:unity, type:others, task:pose landmarker

#5607 - Whether Mediapipe supports the overall detection of multiple people

Issue - State: closed - Opened by jhlibra 17 days ago - 1 comment
Labels: type:feature, platform:c++, stat:awaiting googler, task:holistic landmarker

#5606 - Face Landmark Detection on Stiched Image

Issue - State: open - Opened by aankur 17 days ago - 1 comment
Labels: type:support, platform:javascript, stat:awaiting googler, task:face landmarker

#5605 - Provide bzlmod support

Issue - State: open - Opened by Algomorph 18 days ago - 5 comments
Labels: type:feature, task::all, platform:c++, stat:awaiting googler

#5604 - Desktop examples use the deprecated mediapipe::GlCalculatorHelper::CreateSourceTexture

Issue - State: open - Opened by Algomorph 18 days ago - 2 comments
Labels: platform:c++, type:others

#5603 - An error occurred: module 'mediapipe.python._framework_bindings.model_ckpt_util' has no attribute 'GenerateGpuTfLite'

Issue - State: open - Opened by akau16 18 days ago - 5 comments
Labels: os:windows, platform:python, type:bug, stat:awaiting googler, task:LLM inference

#5602 - Y works

Pull Request - State: closed - Opened by jakan2 18 days ago - 2 comments

#5601 - iOS LLM Task SDK is NOT initialized successfully via KMP

Issue - State: open - Opened by 2BAB 19 days ago - 2 comments
Labels: type:support, platform:ios, task:LLM inference

#5600 - [LLM][falcon_cpu]Failed to run on Android Device (internal:Failed to initialize engine: %s Failed to create engine: INTERNAL:;RET_CHECK failure(mediapipe/tasks/cc/genai/inference/utils/xnn_utils/tflite_weight_accessor.cc:102)tflite_model_

Issue - State: open - Opened by charles-cloud 19 days ago - 2 comments
Labels: type:support, platform:android, task:LLM inference

#5599 - inconsistent face landmarker detection result

Issue - State: closed - Opened by gubinfdu 19 days ago - 6 comments
Labels: type:support, legacy:face mesh, platform:python

#5598 - 如何将mediapipe手部关键点检测的模型导出为.tflite格式

Issue - State: open - Opened by Criminal-9527 21 days ago - 1 comment
Labels: stat:awaiting response, type:others

#5597 - Use local model file but have an error "Not allowed to load local resource: file:///D:/model/gemma-2b-it-gpu-int4.bin"

Issue - State: open - Opened by akau16 22 days ago - 2 comments
Labels: type:support, platform:javascript, task:LLM inference

#5596 - Segmentation fault at TfLiteTensorsToDetectionsCalculator::ProcessCPU

Issue - State: open - Opened by Danil0v3s 23 days ago - 6 comments
Labels: type:support, legacy:autoflip, platform:c++

#5595 - Mediapipe Blendshape

Issue - State: open - Opened by sarperkilic 23 days ago
Labels: type:support, legacy:face mesh, platform:python, os:linux-non-arm

#5594 - Support for GEMMA-2-2B-it Model

Issue - State: open - Opened by jfduma 23 days ago - 6 comments
Labels: type:feature, task:LLM inference

#5593 - hand_tracking_gpu build successfully but can't run

Issue - State: open - Opened by qinweisen 23 days ago - 4 comments
Labels: type:build/install, legacy:hands, platform:c++, os:linux-non-arm

#5592 - Use ImageSegmenter instance as reactive leads to exception

Issue - State: open - Opened by amiznikov 24 days ago
Labels: platform:javascript, type:bug, task:image segmentation, os:macOS

#5591 - Invalid modifications to min_tracking_combinance and min_detection_combinance

Issue - State: open - Opened by wangweikang3-tal 24 days ago - 1 comment
Labels: type:support, platform:c++, task:hand landmarker, os:macOS

#5590 - Gemma 2b converted to tflite seems to produce random results

Issue - State: open - Opened by AmarOk1412 25 days ago - 4 comments
Labels: type:support, platform:android, type:performance, task:text classification

#5589 - Android Image Generation Produces Pure Black

Issue - State: open - Opened by probable1333 25 days ago - 1 comment
Labels: platform:android, type:bug, stat:awaiting googler, task:image generator

#5588 - Inconsistent MediaPipe Tasks Vision version on Maven repository

Issue - State: closed - Opened by ArdeoDeo 25 days ago - 5 comments
Labels: platform:android, task::all, type:others

#5587 - Possibility of calculating facial blendshapes from facial landmarks

Issue - State: open - Opened by neutronengas 26 days ago - 1 comment
Labels: type:feature, platform:python, stat:awaiting googler, task:face landmarker, os:linux-non-arm

#5586 - Colab crash while converting gemma model with mediapipe

Issue - State: closed - Opened by AmarOk1412 29 days ago - 4 comments
Labels: platform:python, type:bug, task:LLM inference

#5585 - Draw over mediapipe output using custom shader in Android

Issue - State: open - Opened by pabssen1 29 days ago
Labels: type:support, platform:android, legacy:object detection

#5584 - Selective RegionOfInterest isn't available in FaceLandmarker task

Issue - State: open - Opened by RandomUserUsingGitHub 29 days ago
Labels: type:support, os:windows, platform:javascript, task:face landmarker

#5583 - Error: NOT_FOUND: Input tensor has type float32: it requires specifying NormalizationOptions metadata to preprocess input images.; Initialize was not ok; StartGraph failed

Issue - State: open - Opened by wanggoldx 29 days ago - 2 comments
Labels: type:support, platform:python, task:image classification

#5582 - How can I run the face landmark model with attention on lips and eyes to provide more accuracy, and additionally output iris landmarks

Issue - State: open - Opened by jhlibra 29 days ago - 1 comment
Labels: type:support, os:windows, legacy:holistic, platform:c++

#5581 - Customize the category of the track

Issue - State: closed - Opened by tisaud 29 days ago - 4 comments
Labels: type:support, platform:android, platform:c++, legacy: box tracking

#5580 - Failed to set name for thread: mediapipe when running invoking endpoint on sagemaker

Issue - State: open - Opened by Manishthakur2503 30 days ago
Labels: type:build/install, platform:python, platform:docker, os:linux-non-arm

#5577 - <ValueError: Expected a model with 4 output tensors, found 2> when trying to instantiate mp.tasks.vision.ObjectDetector in Python with recommended model

Issue - State: closed - Opened by Callum-Halton about 1 month ago - 7 comments
Labels: type:support, platform:python, task:object detection, os:macOS

#5576 - Holistic task error when setting outputFaceBlendshapes to true

Issue - State: open - Opened by amokto about 1 month ago - 2 comments
Labels: type:support, platform:javascript, task:holistic landmarker

#5574 - Duplicate class

Issue - State: closed - Opened by cgowthamanmca about 1 month ago - 6 comments
Labels: type:support, platform:android, legacy:face mesh, legacy:pose

#5571 - Hand World Landmarks are not accurate

Issue - State: open - Opened by Javacr about 1 month ago - 4 comments
Labels: os:windows, platform:python, type:performance, type:bug, stat:awaiting googler, task:gesture recognition

#5570 - Conversion of gemma-2-2b-it model to TensorFlow Lite

Issue - State: open - Opened by shubham0204 about 1 month ago - 12 comments
Labels: type:feature, platform:python, stat:awaiting googler, os:linux-non-arm, task:LLM inference

#5566 - Added iOS refactored LlmInference, LlmTaskRunner and LlmSessionRunner

Pull Request - State: closed - Opened by priankakariatyml about 1 month ago - 1 comment

#5564 - [LLM][Android]SIGSEGV in libllm_inference_engine_jni.so (offset 0x8b0000) ((anonymous namespace)::start_llm_function

Issue - State: closed - Opened by charles-cloud about 1 month ago - 9 comments
Labels: type:support, platform:android, task:LLM inference

#2430 - error:could not con ＇result' from 'std: :unique_ptr <nediapipe: : {anonymous}::GlProcessor,std::default_delete<mediapi ｛anonymous} : : GLProcessor> >

Issue - State: closed - Opened by only-tao about 3 years ago - 10 comments
Labels: type:build/install, platform:python

GitHub / google-ai-edge/mediapipe issues and pull requests