{"id":23273506,"url":"https://github.com/afondiel/Edge-AI-Model-Zoo","last_synced_at":"2025-08-21T05:31:52.390Z","repository":{"id":266547165,"uuid":"898642155","full_name":"afondiel/Edge-AI-Model-Zoo","owner":"afondiel","description":"A list of production-ready models for resource-constrained devices.","archived":false,"fork":false,"pushed_at":"2025-07-12T12:35:57.000Z","size":935,"stargazers_count":4,"open_issues_count":0,"forks_count":2,"subscribers_count":2,"default_branch":"main","last_synced_at":"2025-08-04T06:54:40.629Z","etag":null,"topics":["ai-acceleration","aiot","edge-ai","edge-ai-models","edge-computing","edge-devices","edge-impulse","edge-model-zoo","embedded-ai","executorch","iot","litert","mcus","on-device-models","openvino","optimized-models","qualcomm","ready-to-deploy-models","research-to-production-models","tensorrt"],"latest_commit_sha":null,"homepage":"","language":null,"has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"apache-2.0","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/afondiel.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":"CONTRIBUTING.md","funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null}},"created_at":"2024-12-04T19:01:30.000Z","updated_at":"2025-07-21T09:07:29.000Z","dependencies_parsed_at":"2024-12-04T20:37:08.611Z","dependency_job_id":"53fd9c0c-deb3-431e-8d4d-1ce64a989c49","html_url":"https://github.com/afondiel/Edge-AI-Model-Zoo","commit_stats":null,"previous_names":["afondiel/edgeai-model-zoo","afondiel/edge-ai-model-zoo"],"tags_count":0,"template":false,"template_full_name":null,"purl":"pkg:github/afondiel/Edge-AI-Model-Zoo","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/afondiel%2FEdge-AI-Model-Zoo","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/afondiel%2FEdge-AI-Model-Zoo/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/afondiel%2FEdge-AI-Model-Zoo/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/afondiel%2FEdge-AI-Model-Zoo/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/afondiel","download_url":"https://codeload.github.com/afondiel/Edge-AI-Model-Zoo/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/afondiel%2FEdge-AI-Model-Zoo/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":271430746,"owners_count":24758363,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2022-07-04T15:15:14.044Z","status":"online","status_checked_at":"2025-08-21T02:00:08.990Z","response_time":74,"last_error":null,"robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":true,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["ai-acceleration","aiot","edge-ai","edge-ai-models","edge-computing","edge-devices","edge-impulse","edge-model-zoo","embedded-ai","executorch","iot","litert","mcus","on-device-models","openvino","optimized-models","qualcomm","ready-to-deploy-models","research-to-production-models","tensorrt"],"created_at":"2024-12-19T20:00:32.791Z","updated_at":"2025-08-21T05:31:52.379Z","avatar_url":"https://github.com/afondiel.png","language":null,"funding_links":[],"categories":[],"sub_categories":[],"readme":"[![](https://img.shields.io/badge/Contribute-Welcome-green)](./CONTRIBUTING.md)\n\n# Edge-AI Model Zoos\n\nA curated list of Model Zoos \u0026 Hubs where you can find production-ready and optimized models for resource-constrained devices.\n\n## Table of Contents\n1. [Model Zoos \u0026 Hubs](#1-model-zoos--hubs)\n2. [Model by Domain \u0026 Use Case](#2-model-by-domain--use-case)\n3. [Resources](#resources)\n    - [How to choose the best model for Edge AI application](#how-to-choose-the-best-model-for-edge-ai-application)\n    - [Edge AI Technical Guide for Developers and Practitioners](#edge-ai-technical-guide-for-developers-and-practitioners)\n\n## 1. Model Zoos \u0026 Hubs\n\n[Back to Table of Contents](#table-of-contents)\n\n| Model Zoo | Description | Links |\n|------------------------|---------------------------------------------------------------------|----------------------------------------------------|\n| Edge AI Labs Model Zoo | A collection of pre-trained, optimized models for low-power devices.| [EdgeAI Labs](https://edgeai.modelnova.ai/models/) |\n| Edge Impulse Model Zoo | A repository of models optimized for edge devices. | [Edge Impulse Model Zoo](https://www.edgeimpulse.com/) |\n| ONNX Model Zoo | A collection of pre-trained, state-of-the-art models in the ONNX format. | [ONNX Model Zoo](https://github.com/onnx/models) |\n| NVIDIA Pretrained AI Models (NGC + TAO)| Accelerate AI development with world-class customizable pretrained models from NVIDIA. | - [NVIDIA Pretrained AI Models - Main](https://developer.nvidia.com/ai-models) \u003cbr\u003e - [NGC Model Catalog](https://catalog.ngc.nvidia.com/models?filters=\u0026orderBy=weightPopularDESC\u0026query=\u0026page=\u0026pageSize=) \u003cbr\u003e -  [TAO Model Zoo](https://docs.nvidia.com/tao/tao-toolkit/text/model_zoo/overview.html)|\n| OpenVINO Model Zoo | A collection of pre-trained models ready for use with Intel's OpenVINO toolkit. | [OpenVINO Model Zoo](https://github.com/openvinotoolkit/open_model_zoo) |\n| Qualcomm Models Zoo | A collection of AI models from Qualcomm. | [Qualcomm Models Zoo](https://github.com/quic/ai-hub-models/) |\n| LiteRT Pre-trained models | Pre-trained models optimized for Google's Lite Runtime. | [LiteRT Pre-trained Models](https://ai.google.dev/edge/litert/models/trained) |\n| Keras Applications | Pre-trained models for Keras applications| [Keras Pre-trained Models](https://keras.io/api/applications/#available-models) |\n| MediaPipe | Framework for building multimodal applied machine learning pipelines. | [MediaPipe](https://ai.google.dev/edge/mediapipe/solutions/guide) |\n| TensorFlow Model Garden | A repository with a collection of TensorFlow models. | [TensorFlow Model Garden](https://github.com/tensorflow/models/tree/master) |\n| Pytorch Model Zoo | A hub for pre-trained models on PyTorch framework. | [Pytorch Model Zoo](https://pytorch.org/serve/model_zoo.html) |\n| stm32ai-modelzoo | AI Model Zoo for STM32 microcontroller devices. | [stm32ai-modelzoo](https://github.com/STMicroelectronics/stm32ai-modelzoo/) |\n| Model Zoo | A collection of pre-trained models for various machine learning tasks. | [Model Zoo](https://modelzoo.co/) |\n| Hugging Face Models| A collection of pre-trained models for various machine learning tasks. | [Hugging Face Models](https://huggingface.co/models) |\n| Papers with Code | A repository that links academic papers to their respective code and models. | [Papers with Code](https://paperswithcode.com/) |\n| MXNet Model Zoo | A collection of pre-trained models for the Apache MXNet framework. | [MXNet Model Zoo](https://mxnet.apache.org/versions/1.1.0/model_zoo/index.html) |\n| Deci’s Model Zoo | A curated list of high-performance deep learning models. | [Deci’s Model Zoo](https://deci.ai/modelzoo/) |\n| Jetson Model Zoo and Community Projects | NVIDIA's collection of models and projects for Jetson platform. | [Jetson Model Zoo and Community Projects](https://developer.nvidia.com/embedded/community/jetson-projects) |\n| Magenta | Models for music and art generation from Google's Magenta project. | [Magenta](https://github.com/magenta/magenta/tree/main/magenta/models/arbitrary_image_stylization) |\n| Awesome-CoreML-Models Public | A collection of CoreML models for iOS developers. | [Awesome-CoreML-Models Public](https://github.com/likedan/Awesome-CoreML-Models) |\n| Pinto Models | A variety of models for computer vision tasks. | [Pinto Models](https://github.com/PINTO0309/PINTO_model_zoo) |\n| Baidu AI Open Model Zoo | Baidu's collection of AI models. | [Baidu AI Open Model Zoo](https://ai.baidu.com/tech/modelzoo) |\n| Hailo Model Zoo | A set of models optimized for Hailo's AI processors. | [Hailo Model Zoo](https://github.com/hailo-ai/hailo_model_zoo) |\n\n## 2. Model by Domain \u0026 Use Case\n\n[Back to Table of Contents](#table-of-contents)\n\nThis is a non-exhaustive selection of models from several platforms listed in [Section 1](#1-model-zoos--hubs), ranged into six domains and a variety of tasks, with a focus on efficiency and real-world applications.\n\n\u003ctable border=\"1\"\u003e\n  \u003cthead\u003e\n    \u003ctr\u003e\n      \u003cth\u003eDomain\u003c/th\u003e\n      \u003cth\u003eTask\u003c/th\u003e\n      \u003cth\u003eModel\u003c/th\u003e\n      \u003cth\u003eDescription\u003c/th\u003e\n      \u003cth\u003eReference\u003c/th\u003e\n    \u003c/tr\u003e\n  \u003c/thead\u003e\n  \u003ctbody\u003e\n    \u003c!-- Computer Vision: 10 rows --\u003e\n    \u003ctr\u003e\n      \u003ctd rowspan=\"10\"\u003eComputer Vision\u003c/td\u003e\n      \u003ctd\u003eObject Detection\u003c/td\u003e\n      \u003ctd\u003eyolov8_det\u003c/td\u003e\n      \u003ctd\u003eObject detection for edge devices\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://github.com/ultralytics/yolov8\"\u003eYOLOv8 on GitHub\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eImage Classification\u003c/td\u003e\n      \u003ctd\u003emobilenet_v3_small\u003c/td\u003e\n      \u003ctd\u003eLightweight image classification\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://tfhub.dev/google/imagenet/mobilenet_v3_small_100_224/classification/5\"\u003eMobileNetV3 on TensorFlow Hub\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eSemantic Segmentation\u003c/td\u003e\n      \u003ctd\u003edeeplabv3_resnet50\u003c/td\u003e\n      \u003ctd\u003eSemantic image segmentation\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://tfhub.dev/tensorflow/deeplabv3/1\"\u003eDeepLabV3 on TensorFlow Hub\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eInstance Segmentation\u003c/td\u003e\n      \u003ctd\u003eyolov8_seg\u003c/td\u003e\n      \u003ctd\u003eObject detection and segmentation\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://github.com/ultralytics/yolov8\"\u003eYOLOv8 on GitHub\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eObject Tracking\u003c/td\u003e\n      \u003ctd\u003eDeepSort\u003c/td\u003e\n      \u003ctd\u003eReal-time object tracking\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://github.com/nwojke/deep_sort\"\u003eDeepSort on GitHub\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003ePose Estimation\u003c/td\u003e\n      \u003ctd\u003eopenpose\u003c/td\u003e\n      \u003ctd\u003eHuman pose estimation\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://github.com/CMU-Perceptual-Computing-Lab/openpose\"\u003eOpenPose on GitHub\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eFacial Recognition\u003c/td\u003e\n      \u003ctd\u003emediapipe_face\u003c/td\u003e\n      \u003ctd\u003eFace detection and recognition\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://ai.googleblog.com/2019/08/on-device-real-time-hand-tracking-with.html\"\u003eMediaPipe Face on Google AI\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eOptical Character Recognition\u003c/td\u003e\n      \u003ctd\u003etrocr\u003c/td\u003e\n      \u003ctd\u003eText recognition in images\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://huggingface.co/microsoft/trocr-base\"\u003eTrOCR on Hugging Face\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eVideo Classification\u003c/td\u003e\n      \u003ctd\u003eresnet_2plus1d\u003c/td\u003e\n      \u003ctd\u003eVideo classification for action recognition\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://pytorch.org/hub/facebookresearch_pytorchvideo_resnet/\"\u003eResNet-2+1D on PyTorch Hub\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eVideo Classification\u003c/td\u003e\n      \u003ctd\u003eresnet_3d\u003c/td\u003e\n      \u003ctd\u003e3D CNN for video classification\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://pytorch.org/hub/facebookresearch_pytorchvideo_resnet/\"\u003eResNet-3D on PyTorch Hub\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003c!-- Audio Processing: 10 rows --\u003e\n    \u003ctr\u003e\n      \u003ctd rowspan=\"10\"\u003eAudio Processing\u003c/td\u003e\n      \u003ctd\u003eSpeech-to-Text\u003c/td\u003e\n      \u003ctd\u003edistil-whisper\u003c/td\u003e\n      \u003ctd\u003eLightweight speech recognition model\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://huggingface.co/distil-whisper/distil-large-v3\"\u003eDistil-Whisper on Hugging Face\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eSound Classification\u003c/td\u003e\n      \u003ctd\u003eaudio-spectrogram-transformer\u003c/td\u003e\n      \u003ctd\u003eTransformer for audio classification\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://huggingface.co/MIT/ast-finetuned-audioset\"\u003eAST on Hugging Face\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eVoice Activity Detection\u003c/td\u003e\n      \u003ctd\u003esilero-vad\u003c/td\u003e\n      \u003ctd\u003eVoice activity detection for edge devices\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://github.com/snakers4/silero-vad\"\u003eSilero VAD on GitHub\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eAcoustic Scene Classification\u003c/td\u003e\n      \u003ctd\u003epanns\u003c/td\u003e\n      \u003ctd\u003eAudio tagging and scene classification\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://github.com/qiuqiangkong/panns\"\u003ePANNS on GitHub\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eSpeaker Diarization\u003c/td\u003e\n      \u003ctd\u003epyannote-audio\u003c/td\u003e\n      \u003ctd\u003eSpeaker diarization and segmentation\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://huggingface.co/pyannote/speaker-diarization\"\u003ePyAnnote on Hugging Face\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eSpeech Recognition\u003c/td\u003e\n      \u003ctd\u003ewav2vec2\u003c/td\u003e\n      \u003ctd\u003eSelf-supervised speech representation learning\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://huggingface.co/facebook/wav2vec2-base-960h\"\u003eWav2vec2 on Hugging Face\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003c!-- Time Series: 10 rows --\u003e\n    \u003ctr\u003e\n      \u003ctd rowspan=\"10\"\u003eTime Series\u003c/td\u003e\n      \u003ctd\u003ePredictive Maintenance\u003c/td\u003e\n      \u003ctd\u003etsmixer\u003c/td\u003e\n      \u003ctd\u003eTime-series forecasting for maintenance\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://github.com/google-research/timesfm\"\u003eTimesFM on GitHub\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eAnomaly Detection\u003c/td\u003e\n      \u003ctd\u003eIsolationForest\u003c/td\u003e\n      \u003ctd\u003eAnomaly detection in time-series data\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://scikit-learn.org/stable/modules/generated/sklearn.ensemble.IsolationForest.html\"\u003eIsolationForest on Scikit-learn\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eForecasting\u003c/td\u003e\n      \u003ctd\u003einformer\u003c/td\u003e\n      \u003ctd\u003eTransformer-based time-series forecasting\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://github.com/zhouhaoyi/Informer2020\"\u003eInformer on GitHub\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eTime-Series Classification\u003c/td\u003e\n      \u003ctd\u003erocket\u003c/td\u003e\n      \u003ctd\u003eEfficient time-series classification\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://github.com/angus924/rocket\"\u003eROCKET on GitHub\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eImage Super-Resolution\u003c/td\u003e\n      \u003ctd\u003ereal_esrgan_x4plus\u003c/td\u003e\n      \u003ctd\u003eImage super-resolution for temporal data\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://github.com/xinntao/Real-ESRGAN\"\u003eReal-ESRGAN on GitHub\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eImage Inpainting\u003c/td\u003e\n      \u003ctd\u003elama_dilated\u003c/td\u003e\n      \u003ctd\u003eImage inpainting for time-series analysis\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://github.com/saic-mdal/lama\"\u003eLaMa on GitHub\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003c!-- NLP: 10 rows --\u003e\n    \u003ctr\u003e\n      \u003ctd rowspan=\"10\"\u003eNLP\u003c/td\u003e\n      \u003ctd\u003eSpeech Recognition\u003c/td\u003e\n      \u003ctd\u003eWhisper\u003c/td\u003e\n      \u003ctd\u003eGeneral-purpose speech recognition model\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://huggingface.co/openai/whisper\"\u003eWhisper on Hugging Face\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eKeyword Spotting\u003c/td\u003e\n      \u003ctd\u003esilero-kws\u003c/td\u003e\n      \u003ctd\u003eWake word detection for edge devices\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://github.com/snakers4/silero-models\"\u003eSilero Models on GitHub\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eText Classification\u003c/td\u003e\n      \u003ctd\u003edistilbert\u003c/td\u003e\n      \u003ctd\u003eLightweight transformer for text classification\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://huggingface.co/distilbert-base-uncased\"\u003eDistilBERT on Hugging Face\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eNamed Entity Recognition\u003c/td\u003e\n      \u003ctd\u003ebert-ner\u003c/td\u003e\n      \u003ctd\u003eNER for entity extraction\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://huggingface.co/dslim/bert-base-NER\"\u003eBERT-NER on Hugging Face\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eQuestion Answering\u003c/td\u003e\n      \u003ctd\u003emobilebert\u003c/td\u003e\n      \u003ctd\u003eLightweight QA model for edge\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://huggingface.co/google/mobilebert-uncased\"\u003eMobileBERT on Hugging Face\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eText Summarization\u003c/td\u003e\n      \u003ctd\u003ebart\u003c/td\u003e\n      \u003ctd\u003eText summarization for short texts\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://huggingface.co/facebook/bart-base\"\u003eBART on Hugging Face\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003c!-- Generative AI: Image Generation \u0026 Synthesis (7 rows) --\u003e\n    \u003ctr\u003e\n      \u003ctd rowspan=\"21\"\u003eGenerative AI\u003c/td\u003e\n      \u003ctd rowspan=\"7\"\u003eImage Generation \u0026 Synthesis\u003c/td\u003e\n      \u003ctd\u003eControlNet\u003c/td\u003e\n      \u003ctd\u003eFine control over image generation\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://github.com/lllyasviel/ControlNet\"\u003eControlNet on GitHub\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eStable Diffusion\u003c/td\u003e\n      \u003ctd\u003eText-to-image generation\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://huggingface.co/CompVis/stable-diffusion-v-1-4\"\u003eStable Diffusion on Hugging Face\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003estylegan2\u003c/td\u003e\n      \u003ctd\u003eImage generation\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://github.com/NVlabs/stylegan2\"\u003eStyleGAN2 on GitHub\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eFlux.1-schnell\u003c/td\u003e\n      \u003ctd\u003eFast text-to-image generation\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://huggingface.co/black-forest-labs/FLUX.1-schnell\"\u003eFlux.1 on Hugging Face\u003c/a\u003e, \u003ca href=\"https://github.com/afondiel/awesome-smol\"\u003eAwesome-Smol\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eReve\u003c/td\u003e\n      \u003ctd\u003eImage generation with advanced text rendering\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://huggingface.co/reve/reve\"\u003eReve on Hugging Face\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003c!-- Generative AI: Small LLM (7 rows) --\u003e\n    \u003ctr\u003e\n      \u003ctd rowspan=\"7\"\u003eSmall Language Model (SLM)\u003c/td\u003e\n      \u003ctd\u003eSmolLM2-1.7B\u003c/td\u003e\n      \u003ctd\u003eSmall language model for efficient text generation\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://huggingface.co/HuggingFaceTB/SmolLM2-1.7B\"\u003eSmolLM2 on Hugging Face\u003c/a\u003e, \u003ca href=\"https://github.com/afondiel/awesome-smol\"\u003eAwesome-Smol\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eGemma 2\u003c/td\u003e\n      \u003ctd\u003eLightweight open model for text generation\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://huggingface.co/google/gemma-2-2b\"\u003eGemma 2 on Hugging Face\u003c/a\u003e, \u003ca href=\"https://github.com/afondiel/awesome-smol\"\u003eAwesome-Smol\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003ePhi-3.5-mini\u003c/td\u003e\n      \u003ctd\u003eSmall language model with strong reasoning\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://huggingface.co/microsoft/phi-3.5-mini-instruct\"\u003ePhi-3.5-mini on Hugging Face\u003c/a\u003e, \u003ca href=\"https://github.com/afondiel/awesome-smol\"\u003eAwesome-Smol\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eQwen2.5-1.5B\u003c/td\u003e\n      \u003ctd\u003eEfficient language model for instruction following\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://huggingface.co/Qwen/Qwen2.5-1.5B\"\u003eQwen2.5 on Hugging Face\u003c/a\u003e, \u003ca href=\"https://github.com/afondiel/awesome-smol\"\u003eAwesome-Smol\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eMixtral-8x22B\u003c/td\u003e\n      \u003ctd\u003eSparse mixture of experts for text generation\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://huggingface.co/mixtral-8x22b\"\u003eMixtral on Hugging Face\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003c!-- Generative AI: Multimodality (7 rows) --\u003e\n    \u003ctr\u003e\n      \u003ctd rowspan=\"7\"\u003eMultimodality\u003c/td\u003e\n      \u003ctd\u003eSmolVLM-256M\u003c/td\u003e\n      \u003ctd\u003eSmallest vision-language model for image understanding\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://huggingface.co/HuggingFaceTB/SmolVLM-256M\"\u003eSmolVLM-256M on Hugging Face\u003c/a\u003e, \u003ca href=\"https://github.com/afondiel/awesome-smol\"\u003eAwesome-Smol\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eSmolVLM-500M\u003c/td\u003e\n      \u003ctd\u003eVision-language model for image and text tasks\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://huggingface.co/HuggingFaceTB/SmolVLM-500M\"\u003eSmolVLM-500M on Hugging Face\u003c/a\u003e, \u003ca href=\"https://github.com/afondiel/awesome-smol\"\u003eAwesome-Smol\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eBakLLaVA-1\u003c/td\u003e\n      \u003ctd\u003eMultimodal model for text and image tasks\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://huggingface.co/SkunkworksAI/BakLLaVA-1\"\u003eBakLLaVA-1 on Hugging Face\u003c/a\u003e, \u003ca href=\"https://github.com/afondiel/awesome-smol\"\u003eAwesome-Smol\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003ePaliGemma\u003c/td\u003e\n      \u003ctd\u003eVision-language model for multimodal tasks\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://huggingface.co/google/paligemma\"\u003ePaliGemma on Hugging Face\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eSeed1.5-VL\u003c/td\u003e\n      \u003ctd\u003eVision-language model with strong multimodal performance\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://huggingface.co/ByteDance/Seed1.5-VL\"\u003eSeed1.5-VL on Hugging Face\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003c!-- Misc: 10 rows --\u003e\n    \u003ctr\u003e\n      \u003ctd rowspan=\"10\"\u003eMisc\u003c/td\u003e\n      \u003ctd\u003eSensor Fusion\u003c/td\u003e\n      \u003ctd\u003emediapipe_pose\u003c/td\u003e\n      \u003ctd\u003eHuman pose estimation using sensor data\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://ai.googleblog.com/2019/08/on-device-real-time-hand-tracking-with.html\"\u003eMediaPipe Pose on Google AI\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eActivity Recognition\u003c/td\u003e\n      \u003ctd\u003ehar-cnn\u003c/td\u003e\n      \u003ctd\u003eHuman activity recognition from sensor data\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://github.com/saif-mahmud/human-activity-recognition\"\u003eHAR-CNN on GitHub\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eContextual Awareness\u003c/td\u003e\n      \u003ctd\u003eSmolVLM-256M\u003c/td\u003e\n      \u003ctd\u003eMultimodal model for environment understanding\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://huggingface.co/HuggingFaceTB/SmolVLM-256M\"\u003eSmolVLM-256M on Hugging Face\u003c/a\u003e, \u003ca href=\"https://github.com/afondiel/awesome-smol\"\u003eAwesome-Smol\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eNetwork Anomaly Detection\u003c/td\u003e\n      \u003ctd\u003eLOF\u003c/td\u003e\n      \u003ctd\u003eLocal outlier factor for network anomalies\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://scikit-learn.org/stable/modules/generated/sklearn.neighbors.LocalOutlierFactor.html\"\u003eLOF on Scikit-learn\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eDevice Behavior Anomaly\u003c/td\u003e\n      \u003ctd\u003eAutoencoder\u003c/td\u003e\n      \u003ctd\u003eAnomaly detection for device behavior\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://github.com/keras-team/keras-io/blob/master/examples/timeseries/timeseries_anomaly_detection.py\"\u003eKeras Autoencoder\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eSensor Data Anomaly\u003c/td\u003e\n      \u003ctd\u003eOC-SVM\u003c/td\u003e\n      \u003ctd\u003eOne-class SVM for sensor data anomalies\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://scikit-learn.org/stable/modules/generated/sklearn.svm.OneClassSVM.html\"\u003eOneClassSVM on Scikit-learn\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eOn-device Control Systems\u003c/td\u003e\n      \u003ctd\u003eTD3\u003c/td\u003e\n      \u003ctd\u003eTwin Delayed DDPG for control systems\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://github.com/sfujim/TD3\"\u003eTD3 on GitHub\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eVarious\u003c/td\u003e\n      \u003ctd\u003epinecone\u003c/td\u003e\n      \u003ctd\u003eVector database\u003c/td\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eVarious\u003c/td\u003e\n      \u003ctd\u003eweaviate-c2\u003c/td\u003e\n      \u003ctd\u003eVector database\u003c/td\u003e\n      \u003ctd\u003e-\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eVarious\u003c/td\u003e\n      \u003ctd\u003eupstage\u003c/td\u003e\n      \u003ctd\u003eVarious models\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://github.com/afondiel/awesome-smol\"\u003eAwesome-Smol\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n  \u003c/tbody\u003e\n\u003c/table\u003e\n\n## Resources\n\n### How to choose the best model for an Edge AI application\n\nSelecting the right model for edge deployment is critical for balancing **performance**, **accuracy** and **efficiency**.\n\n### Why It Matters\n- **Efficiency**: Edge devices (e.g., IoT, mobile, embedded systems) have limited compute, memory, and power.\n- **Performance**: Real-time applications (e.g., autonomous drones, smart cameras) demand low latency and high accuracy.\n- **Scalability**: The right model ensures cost-effective deployment across devices.\n\n### Key Criteria\n1. **Task Requirements**: Match the model to your application (e.g., vision, audio, multimodal).\n2. **Hardware Constraints**: Consider compute (OPS), memory (MB), and energy (mWh) limits of your device.\n3. **Performance Goals**: Balance accuracy, latency, and throughput for your use case.\n4. **Deployment Ease**: Check compatibility with frameworks (e.g., TensorFlow Lite, ONNX).\n\n**Next Steps**: Once you’ve shortlisted a model, use the [Edge AI Benchmarking Guide](https://github.com/afondiel/Edge-AI-Benchmarking) to **profile** and **optimize** the model performance.\n\n### Edge AI Technical Guide for Developers and Practitioners\n\n- [Edge AI Engineering](https://github.com/afondiel/edge-ai-engineering) \n- [Edge AI Technical Guide](https://github.com/afondiel/computer-science-notebook/tree/master/core/systems/edge-computing/edge-ai/concepts)\n- [Edge AI End-to-End Stack](https://www.qualcomm.com/developer/artificial-intelligence)\n- [Edge AI Deployment Stack](https://github.com/afondiel/computer-science-notebook/tree/master/core/systems/edge-computing/edge-ai/concepts/deployment)\n- [Edge AI Optimization Stack](https://github.com/afondiel/computer-science-notebook/tree/master/core/systems/edge-computing/edge-ai/concepts/optimization)\n- [Edge AI Frameworks](https://github.com/afondiel/computer-science-notebook/tree/master/core/systems/edge-computing/edge-ai/frameworks)\n- [Edge AI Model Zoos](https://github.com/afondiel/Edge-AI-Model-Zoo)\n- [Edge AI Platforms](https://github.com/afondiel/Edge-AI-Platforms)\n- [Edge AI Benchmarking](https://github.com/afondiel/Edge-AI-Benchmarking)\n- [Edge AI Ecosystem](https://github.com/afondiel/computer-science-notebook/tree/master/core/systems/edge-computing/edge-ai/industry-applications)\n- [Edge AI Books](https://github.com/afondiel/cs-books/blob/main/README.md#edge-computing)\n- [Edge AI Blog](https://afondiel.github.io/posts/)\n- [Edge AI Papers](https://github.com/afondiel/computer-science-notebook/tree/master/core/systems/edge-computing/edge-ai/resources/edge_ai_papers_news.md)\n\n[Back to Table of Contents](#table-of-contents)\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fafondiel%2FEdge-AI-Model-Zoo","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fafondiel%2FEdge-AI-Model-Zoo","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fafondiel%2FEdge-AI-Model-Zoo/lists"}