{"id":17236615,"url":"https://github.com/illright/draw-ill-help","last_synced_at":"2026-07-09T15:30:19.932Z","repository":{"id":45081454,"uuid":"436989961","full_name":"illright/draw-ill-help","owner":"illright","description":"A drawing app with shape correction powered by ML","archived":true,"fork":false,"pushed_at":"2023-10-08T19:35:09.000Z","size":7592,"stargazers_count":18,"open_issues_count":13,"forks_count":2,"subscribers_count":3,"default_branch":"main","last_synced_at":"2025-02-12T13:47:22.654Z","etag":null,"topics":["feature-sliced","svelte","sveltekit","tensorflowjs","typescript","yolo"],"latest_commit_sha":null,"homepage":"https://illright.github.io/draw-ill-help/","language":"TypeScript","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"mit","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/illright.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null}},"created_at":"2021-12-10T13:32:24.000Z","updated_at":"2024-07-27T17:11:01.000Z","dependencies_parsed_at":"2023-02-10T00:00:34.351Z","dependency_job_id":null,"html_url":"https://github.com/illright/draw-ill-help","commit_stats":null,"previous_names":[],"tags_count":1,"template":false,"template_full_name":null,"repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/illright%2Fdraw-ill-help","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/illright%2Fdraw-ill-help/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/illright%2Fdraw-ill-help/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/illright%2Fdraw-ill-help/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/illright","download_url":"https://codeload.github.com/illright/draw-ill-help/tar.gz/refs/heads/main","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":240494695,"owners_count":19810476,"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","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":["feature-sliced","svelte","sveltekit","tensorflowjs","typescript","yolo"],"created_at":"2024-10-15T05:36:33.370Z","updated_at":"2026-07-09T15:30:19.884Z","avatar_url":"https://github.com/illright.png","language":"TypeScript","funding_links":[],"categories":[],"sub_categories":[],"readme":"# Draw, I'll Help\n\nA progressive web application for drawing that automatically corrects hand-drawn circles and rectangles with the power of YOLOv5.\n\nCheck it out at \u003chttps://illright.github.io/draw-ill-help\u003e.\n\n## Highlights\n\n- 🏎️ Completely serverless\n- 🌐 Works offline\n- 🧠 Runs YOLOv5 Nano in the background\n- 🗃️ Built-in dataset generator\n- 😎 Automatic dark mode\n- 🍰 Follows the principles of [Feature-Sliced Design](https://feature-sliced.design/)\n\n## Screenshots\n\n\u003ctable\u003e\n\u003cthead\u003e\n\u003cth\u003eHome page\u003c/th\u003e\n\u003cth\u003eDrawing canvas\u003c/th\u003e\n\u003cth\u003eDataset generator\u003c/th\u003e\n\u003c/thead\u003e\n\u003ctr\u003e\n\u003ctd\u003e\n\n![Home page](./.github/readme/home.jpg)\n\n\u003c/td\u003e\n\u003ctd\u003e\n\n![Drawing page](./.github/readme/draw.jpg)\n\n\u003c/td\u003e\n\u003ctd\u003e\n\n![Dataset page](./.github/readme/dataset.jpg)\n\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003c/table\u003e\n\n## Model performance\n\nSince the selected model is the smallest in the family of YOLOv5, it has very small inference times and file size. However, that also has an impact on its predictive capabilities.\n\nSee the charts below for the main performance characteristics of the model during training. The model had trained for around 350 epochs, early stopping with a patience value of 100 epochs due to no improvement at the target metric (0.9 of **mAP (0.5:0.95)** and 0.1 of **mAP (0.5)**). The horizontal axis represents the epochs (aka steps).\n\n\u003ctable\u003e\n\u003cthead\u003e\n\u003cth\u003emAP (0.5)\u003c/th\u003e\n\u003cth\u003emAP (0.5:0.95)\u003c/th\u003e\n\u003cth\u003ePrecision\u003c/th\u003e\n\u003cth\u003eRecall\u003c/th\u003e\n\u003c/thead\u003e\n\u003ctr\u003e\n\u003ctd\u003e\n\n\u003cpicture\u003e\n  \u003csource srcset=\"./.github/readme/metrics_mAP_0.5.svg\" media=\"(prefers-color-scheme: dark)\" /\u003e\n  \u003cimg\n    width=\"200\"\n    height=\"180\"\n    alt=\"Mean average precision graph, IoU threshold of 0.5. Starting low at 0.3, rising rapidly to 0.9 after 100 steps, staying around 0.93 until the end with a few sharp drops to almost 0.3 at 160, 195, 260 and 300 steps.\"\n    src=\"./.github/readme/metrics_mAP_0.5_light.svg\"\n  /\u003e\n\u003c/picture\u003e\n\n\u003c/td\u003e\n\u003ctd\u003e\n\n\u003cpicture\u003e\n  \u003csource srcset=\"./.github/readme/metrics_mAP_0.5_0.95.svg\" media=\"(prefers-color-scheme: dark)\" /\u003e\n  \u003cimg\n    width=\"200\"\n    height=\"180\"\n    alt=\"Mean average precision graph, averaged across IoU thresholds from 0.5 to 0.95. Starting low at 0.1, rising almost linearly to 0.7 after 100 steps, peaking in 0.89 at step 250 with a few sharp drops to almost 0.1 at 160, 195, 260 and 300 steps.\"\n    src=\"./.github/readme/metrics_mAP_0.5_0.95_light.svg\"\n  /\u003e\n\u003c/picture\u003e\n\n\u003c/td\u003e\n\u003ctd\u003e\n\n\u003cpicture\u003e\n  \u003csource srcset=\"./.github/readme/metrics_precision.svg\" media=\"(prefers-color-scheme: dark)\" /\u003e\n  \u003cimg\n    width=\"200\"\n    height=\"180\"\n    alt=\"Precision graph. Increasing sharply from 0.2 to 0.8 in 20 steps, peaking in 0.92 at step 250 with a few sharp drops at the same steps where mAP had dropped.\"\n    src=\"./.github/readme/metrics_precision_light.svg\"\n  /\u003e\n\u003c/picture\u003e\n\n\u003c/td\u003e\n\u003ctd\u003e\n\n\u003cpicture\u003e\n  \u003csource srcset=\"./.github/readme/metrics_recall.svg\" media=\"(prefers-color-scheme: dark)\" /\u003e\n  \u003cimg\n    width=\"200\"\n    height=\"180\"\n    alt=\"Recall graph. Starting low at 0.2, rising almost rapidly to 0.9 after 50 steps, frequently approaching 1.0, first at step 105, with a few sharp drops at the same steps where mAP had dropped.\"\n    src=\"./.github/readme/metrics_recall_light.svg\"\n  /\u003e\n\u003c/picture\u003e\n\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003c/table\u003e\n\n## Running locally\n\nGrab the `model.zip` file from the latest release and unpack its contents into `static/`. Then run the usual commands:\n\n```bash\npnpm i\npnpm dev\n```\n\n## Training YOLOv5 yourself\n\nThe Jupyter notebook with explanations for the training process is available in the [`yolov5/`](./yolov5) directory.\n\nThe recommended way to run the notebook is with Google Colaboratory. If you're using the GPU session and the free Colab account, the dataset of around 1.5k 416x416 images should take around two hours to train.\n\n[![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/illright/draw-ill-help/blob/main/yolov5/training.ipynb)\n\n## License\n\nThe source code of this project is distributed under the terms of the MIT license. [Click here](https://choosealicense.com/licenses/mit/) to learn what that means.\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fillright%2Fdraw-ill-help","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fillright%2Fdraw-ill-help","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fillright%2Fdraw-ill-help/lists"}