Add files

.gitattributes

@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+*.jpg filter=lfs diff=lfs merge=lfs -text

.pre-commit-config.yaml

@@ -0,0 +1,34 @@
+repos:
+  - repo: https://github.com/pre-commit/pre-commit-hooks
+    rev: v5.0.0
+    hooks:
+      - id: check-executables-have-shebangs
+      - id: check-json
+      - id: check-merge-conflict
+      - id: check-shebang-scripts-are-executable
+      - id: check-toml
+      - id: check-yaml
+      - id: end-of-file-fixer
+      - id: mixed-line-ending
+        args: ["--fix=lf"]
+      - id: requirements-txt-fixer
+      - id: trailing-whitespace
+  - repo: https://github.com/astral-sh/ruff-pre-commit
+    rev: v0.8.6
+    hooks:
+      - id: ruff
+        args: ["--fix"]
+      - id: ruff-format
+        args: ["--line-length", "119"]
+  - repo: https://github.com/pre-commit/mirrors-mypy
+    rev: v1.14.1
+    hooks:
+      - id: mypy
+        args: ["--ignore-missing-imports"]
+        additional_dependencies:
+          [
+            "types-python-slugify",
+            "types-requests",
+            "types-PyYAML",
+            "types-pytz",
+          ]

.python-version

@@ -0,0 +1 @@
+3.10

.vscode/extensions.json

@@ -0,0 +1,8 @@
+{
+    "recommendations": [
+        "ms-python.python",
+        "charliermarsh.ruff",
+        "streetsidesoftware.code-spell-checker",
+        "tamasfe.even-better-toml"
+    ]
+}

.vscode/settings.json

@@ -0,0 +1,17 @@
+{
+    "editor.formatOnSave": true,
+    "files.insertFinalNewline": false,
+    "[python]": {
+        "editor.defaultFormatter": "charliermarsh.ruff",
+        "editor.formatOnType": true,
+        "editor.codeActionsOnSave": {
+            "source.fixAll.ruff": "explicit",
+            "source.organizeImports": "explicit"
+        }
+    },
+    "[jupyter]": {
+        "files.insertFinalNewline": false
+    },
+    "notebook.output.scrolling": true,
+    "notebook.formatOnSave.enabled": true
+}

LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2023 hysts
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

README.md

@@ -1,8 +1,8 @@
 ---
 title: ViTPose Transformers
-emoji: 💻
-colorFrom: indigo
-colorTo: red
+emoji: ⚡
+colorFrom: red
+colorTo: purple
 sdk: gradio
 sdk_version: 5.11.0
 app_file: app.py

app.py

@@ -0,0 +1,127 @@
+#!/usr/bin/env python
+
+"""A demo of the VitPose model.
+
+This code is based on the implementation from the Colab notebook:
+https://colab.research.google.com/drive/1e8fcby5rhKZWcr9LSN8mNbQ0TU4Dxxpo
+"""
+
+import pathlib
+
+import gradio as gr
+import PIL.Image
+import spaces
+import supervision as sv
+import torch
+from transformers import AutoProcessor, RTDetrForObjectDetection, VitPoseForPoseEstimation
+
+DESCRIPTION = "# ViTPose"
+
+device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
+
+person_detector_name = "PekingU/rtdetr_r50vd_coco_o365"
+person_image_processor = AutoProcessor.from_pretrained(person_detector_name)
+person_model = RTDetrForObjectDetection.from_pretrained(person_detector_name, device_map=device)
+
+pose_model_name = "usyd-community/vitpose-base-simple"
+pose_image_processor = AutoProcessor.from_pretrained(pose_model_name)
+pose_model = VitPoseForPoseEstimation.from_pretrained(pose_model_name, device_map=device)
+
+
+@spaces.GPU
+@torch.inference_mode()
+def run(image: PIL.Image.Image) -> tuple[PIL.Image.Image, list[dict]]:
+    inputs = person_image_processor(images=image, return_tensors="pt").to(device)
+    outputs = person_model(**inputs)
+    results = person_image_processor.post_process_object_detection(
+        outputs, target_sizes=torch.tensor([(image.height, image.width)]), threshold=0.3
+    )
+    result = results[0]  # take first image results
+
+    # Human label refers 0 index in COCO dataset
+    person_boxes_xyxy = result["boxes"][result["labels"] == 0]
+    person_boxes_xyxy = person_boxes_xyxy.cpu().numpy()
+
+    # Convert boxes from VOC (x1, y1, x2, y2) to COCO (x1, y1, w, h) format
+    person_boxes = person_boxes_xyxy.copy()
+    person_boxes[:, 2] = person_boxes[:, 2] - person_boxes[:, 0]
+    person_boxes[:, 3] = person_boxes[:, 3] - person_boxes[:, 1]
+
+    inputs = pose_image_processor(image, boxes=[person_boxes], return_tensors="pt").to(device)
+
+    # for vitpose-plus-base checkpoint we should additionaly provide dataset_index
+    # to sepcify which MOE experts to use for inference
+    if pose_model.config.backbone_config.num_experts > 1:
+        dataset_index = torch.tensor([0] * len(inputs["pixel_values"]))
+        dataset_index = dataset_index.to(inputs["pixel_values"].device)
+        inputs["dataset_index"] = dataset_index
+
+    outputs = pose_model(**inputs)
+
+    pose_results = pose_image_processor.post_process_pose_estimation(outputs, boxes=[person_boxes])
+    image_pose_result = pose_results[0]  # results for first image
+
+    # make results more human-readable
+    human_readable_results = []
+    for i, person_pose in enumerate(image_pose_result):
+        data = {
+            "person_id": i,
+            "bbox": person_pose["bbox"].numpy().tolist(),
+            "keypoints": [],
+        }
+        for keypoint, label, score in zip(
+            person_pose["keypoints"], person_pose["labels"], person_pose["scores"], strict=True
+        ):
+            keypoint_name = pose_model.config.id2label[label.item()]
+            x, y = keypoint
+            data["keypoints"].append({"name": keypoint_name, "x": x.item(), "y": y.item(), "score": score.item()})
+        human_readable_results.append(data)
+
+    # preprocess to torch tensor of shape (n_objects, n_keypoints, 2)
+    xy = [pose_result["keypoints"] for pose_result in image_pose_result]
+    xy = torch.stack(xy).cpu().numpy()
+
+    scores = [pose_result["scores"] for pose_result in image_pose_result]
+    scores = torch.stack(scores).cpu().numpy()
+
+    keypoints = sv.KeyPoints(xy=xy, confidence=scores)
+    detections = sv.Detections(xyxy=person_boxes_xyxy)
+
+    edge_annotator = sv.EdgeAnnotator(color=sv.Color.GREEN, thickness=1)
+    vertex_annotator = sv.VertexAnnotator(color=sv.Color.RED, radius=2)
+    bounding_box_annotator = sv.BoxAnnotator(color=sv.Color.WHITE, color_lookup=sv.ColorLookup.INDEX, thickness=1)
+
+    annotated_frame = image.copy()
+
+    # annotate boundg boxes
+    annotated_frame = bounding_box_annotator.annotate(scene=image.copy(), detections=detections)
+
+    # annotate edges and verticies
+    annotated_frame = edge_annotator.annotate(scene=annotated_frame, key_points=keypoints)
+    return vertex_annotator.annotate(scene=annotated_frame, key_points=keypoints), human_readable_results
+
+
+paths = sorted(pathlib.Path("images").glob("*.jpg"))
+
+
+with gr.Blocks(css_paths="style.css") as demo:
+    gr.Markdown(DESCRIPTION)
+    with gr.Row():
+        with gr.Column():
+            input_image = gr.Image(label="Input Image", type="pil")
+            run_button = gr.Button()
+        with gr.Column():
+            output_image = gr.Image(label="Output Image")
+            output_json = gr.JSON(label="Output JSON")
+
+    gr.Examples(examples=paths, inputs=input_image, outputs=[output_image, output_json], fn=run)
+
+    run_button.click(
+        fn=run,
+        inputs=input_image,
+        outputs=[output_image, output_json],
+    )
+
+
+if __name__ == "__main__":
+    demo.queue(max_size=20).launch()

images/README.md

@@ -0,0 +1,9 @@
+These images are from the following public domain:
+
+- https://www.pexels.com/photo/women-in-active-wear-balancing-their-body-while-leaning-by-the-doorway-5770445/
+- https://www.pexels.com/photo/woman-balancing-her-body-on-a-handstand-using-one-hand-5770708/
+- https://www.pexels.com/photo/persons-in-black-shirt-and-pants-690598/
+- https://www.pexels.com/photo/photo-of-woman-doing-a-ballet-dance-1164975/
+- https://www.pexels.com/photo/beautiful-woman-in-a-red-dress-wearing-red-lipstick-7909580/
+- https://www.pexels.com/photo/girl-in-red-jacket-riding-bicycle-5792907/
+- https://www.pexels.com/photo/woman-wearing-a-white-gown-walking-on-grass-field-8574605/

pyproject.toml

@@ -0,0 +1,58 @@
+[project]
+name = "vitpose-transformers"
+version = "0.1.0"
+description = ""
+readme = "README.md"
+requires-python = ">=3.10"
+dependencies = [
+    "accelerate>=1.2.1",
+    "gradio>=5.11.0",
+    "hf-transfer>=0.1.9",
+    "setuptools>=75.8.0",
+    "spaces>=0.32.0",
+    "supervision>=0.25.1",
+    "torch==2.4.0",
+    "transformers>=4.47.1",
+]
+
+[tool.ruff]
+line-length = 119
+
+[tool.ruff.lint]
+select = ["ALL"]
+ignore = [
+    "COM812", # missing-trailing-comma
+    "D203",   # one-blank-line-before-class
+    "D213",   # multi-line-summary-second-line
+    "E501",   # line-too-long
+    "SIM117", # multiple-with-statements
+]
+extend-ignore = [
+    "D100",    # undocumented-public-module
+    "D101",    # undocumented-public-class
+    "D102",    # undocumented-public-method
+    "D103",    # undocumented-public-function
+    "D104",    # undocumented-public-package
+    "D105",    # undocumented-magic-method
+    "D107",    # undocumented-public-init
+    "EM101",   # raw-string-in-exception
+    "FBT001",  # boolean-type-hint-positional-argument
+    "FBT002",  # boolean-default-value-positional-argument
+    "PD901",   # pandas-df-variable-name
+    "PGH003",  # blanket-type-ignore
+    "PLR0913", # too-many-arguments
+    "PLR0915", # too-many-statements
+    "TRY003",  # raise-vanilla-args
+]
+unfixable = [
+    "F401", # unused-import
+]
+
+[tool.ruff.lint.per-file-ignores]
+"*.ipynb" = ["T201"]
+
+[tool.ruff.format]
+docstring-code-format = true
+
+[tool.uv.sources]
+transformers = { git = "https://github.com/huggingface/transformers.git", rev = "a6256ec0982fee2c57cc41237bff7e64ed4dcda9" }

requirements.txt

@@ -0,0 +1,278 @@
+# This file was autogenerated by uv via the following command:
+#    uv pip compile pyproject.toml -o requirements.txt
+accelerate==1.2.1
+    # via vitpose-transformers (pyproject.toml)
+aiofiles==23.2.1
+    # via gradio
+annotated-types==0.7.0
+    # via pydantic
+anyio==4.8.0
+    # via
+    #   gradio
+    #   httpx
+    #   starlette
+certifi==2024.12.14
+    # via
+    #   httpcore
+    #   httpx
+    #   requests
+charset-normalizer==3.4.1
+    # via requests
+click==8.1.8
+    # via
+    #   typer
+    #   uvicorn
+contourpy==1.3.1
+    # via
+    #   matplotlib
+    #   supervision
+cycler==0.12.1
+    # via matplotlib
+defusedxml==0.7.1
+    # via supervision
+exceptiongroup==1.2.2
+    # via anyio
+fastapi==0.115.6
+    # via gradio
+ffmpy==0.5.0
+    # via gradio
+filelock==3.16.1
+    # via
+    #   huggingface-hub
+    #   torch
+    #   transformers
+    #   triton
+fonttools==4.55.3
+    # via matplotlib
+fsspec==2024.12.0
+    # via
+    #   gradio-client
+    #   huggingface-hub
+    #   torch
+gradio==5.11.0
+    # via
+    #   vitpose-transformers (pyproject.toml)
+    #   spaces
+gradio-client==1.5.3
+    # via gradio
+h11==0.14.0
+    # via
+    #   httpcore
+    #   uvicorn
+hf-transfer==0.1.9
+    # via vitpose-transformers (pyproject.toml)
+httpcore==1.0.7
+    # via httpx
+httpx==0.28.1
+    # via
+    #   gradio
+    #   gradio-client
+    #   safehttpx
+    #   spaces
+huggingface-hub==0.27.1
+    # via
+    #   accelerate
+    #   gradio
+    #   gradio-client
+    #   tokenizers
+    #   transformers
+idna==3.10
+    # via
+    #   anyio
+    #   httpx
+    #   requests
+jinja2==3.1.5
+    # via
+    #   gradio
+    #   torch
+kiwisolver==1.4.8
+    # via matplotlib
+markdown-it-py==3.0.0
+    # via rich
+markupsafe==2.1.5
+    # via
+    #   gradio
+    #   jinja2
+matplotlib==3.10.0
+    # via supervision
+mdurl==0.1.2
+    # via markdown-it-py
+mpmath==1.3.0
+    # via sympy
+networkx==3.4.2
+    # via torch
+numpy==2.2.1
+    # via
+    #   accelerate
+    #   contourpy
+    #   gradio
+    #   matplotlib
+    #   opencv-python
+    #   pandas
+    #   scipy
+    #   supervision
+    #   transformers
+nvidia-cublas-cu12==12.1.3.1
+    # via
+    #   nvidia-cudnn-cu12
+    #   nvidia-cusolver-cu12
+    #   torch
+nvidia-cuda-cupti-cu12==12.1.105
+    # via torch
+nvidia-cuda-nvrtc-cu12==12.1.105
+    # via torch
+nvidia-cuda-runtime-cu12==12.1.105
+    # via torch
+nvidia-cudnn-cu12==9.1.0.70
+    # via torch
+nvidia-cufft-cu12==11.0.2.54
+    # via torch
+nvidia-curand-cu12==10.3.2.106
+    # via torch
+nvidia-cusolver-cu12==11.4.5.107
+    # via torch
+nvidia-cusparse-cu12==12.1.0.106
+    # via
+    #   nvidia-cusolver-cu12
+    #   torch
+nvidia-nccl-cu12==2.20.5
+    # via torch
+nvidia-nvjitlink-cu12==12.6.85
+    # via
+    #   nvidia-cusolver-cu12
+    #   nvidia-cusparse-cu12
+nvidia-nvtx-cu12==12.1.105
+    # via torch
+opencv-python==4.10.0.84
+    # via supervision
+orjson==3.10.14
+    # via gradio
+packaging==24.2
+    # via
+    #   accelerate
+    #   gradio
+    #   gradio-client
+    #   huggingface-hub
+    #   matplotlib
+    #   spaces
+    #   transformers
+pandas==2.2.3
+    # via gradio
+pillow==11.1.0
+    # via
+    #   gradio
+    #   matplotlib
+    #   supervision
+psutil==5.9.8
+    # via
+    #   accelerate
+    #   spaces
+pydantic==2.10.4
+    # via
+    #   fastapi
+    #   gradio
+    #   spaces
+pydantic-core==2.27.2
+    # via pydantic
+pydub==0.25.1
+    # via gradio
+pygments==2.19.1
+    # via rich
+pyparsing==3.2.1
+    # via matplotlib
+python-dateutil==2.9.0.post0
+    # via
+    #   matplotlib
+    #   pandas
+python-multipart==0.0.20
+    # via gradio
+pytz==2024.2
+    # via pandas
+pyyaml==6.0.2
+    # via
+    #   accelerate
+    #   gradio
+    #   huggingface-hub
+    #   supervision
+    #   transformers
+regex==2024.11.6
+    # via transformers
+requests==2.32.3
+    # via
+    #   huggingface-hub
+    #   spaces
+    #   supervision
+    #   transformers
+rich==13.9.4
+    # via typer
+ruff==0.8.6
+    # via gradio
+safehttpx==0.1.6
+    # via gradio
+safetensors==0.5.2
+    # via
+    #   accelerate
+    #   transformers
+scipy==1.15.0
+    # via supervision
+semantic-version==2.10.0
+    # via gradio
+setuptools==75.8.0
+    # via vitpose-transformers (pyproject.toml)
+shellingham==1.5.4
+    # via typer
+six==1.17.0
+    # via python-dateutil
+sniffio==1.3.1
+    # via anyio
+spaces==0.32.0
+    # via vitpose-transformers (pyproject.toml)
+starlette==0.41.3
+    # via
+    #   fastapi
+    #   gradio
+supervision==0.25.1
+    # via vitpose-transformers (pyproject.toml)
+sympy==1.13.3
+    # via torch
+tokenizers==0.21.0
+    # via transformers
+tomlkit==0.13.2
+    # via gradio
+torch==2.4.0
+    # via
+    #   vitpose-transformers (pyproject.toml)
+    #   accelerate
+tqdm==4.67.1
+    # via
+    #   huggingface-hub
+    #   supervision
+    #   transformers
+transformers @ git+https://github.com/huggingface/transformers.git@a6256ec0982fee2c57cc41237bff7e64ed4dcda9
+    # via vitpose-transformers (pyproject.toml)
+triton==3.0.0
+    # via torch
+typer==0.15.1
+    # via gradio
+typing-extensions==4.12.2
+    # via
+    #   anyio
+    #   fastapi
+    #   gradio
+    #   gradio-client
+    #   huggingface-hub
+    #   pydantic
+    #   pydantic-core
+    #   rich
+    #   spaces
+    #   torch
+    #   typer
+    #   uvicorn
+tzdata==2024.2
+    # via pandas
+urllib3==2.3.0
+    # via requests
+uvicorn==0.34.0
+    # via gradio
+websockets==14.1
+    # via gradio-client

style.css

@@ -0,0 +1,11 @@
+h1 {
+  text-align: center;
+  display: block;
+}
+
+#duplicate-button {
+  margin: auto;
+  color: #fff;
+  background: #1565c0;
+  border-radius: 100vh;
+}