Upload 21 files

.gitattributes

@@ -33,3 +33,6 @@ 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
+assets/outputs/car.png filter=lfs diff=lfs merge=lfs -text
+assets/outputs/fruits.png filter=lfs diff=lfs merge=lfs -text
+assets/outputs/person.png filter=lfs diff=lfs merge=lfs -text

Dockerfile

@@ -0,0 +1,24 @@
+FROM nvidia/cuda:12.4.1-cudnn-devel-ubuntu22.04
+
+# Set environment variables
+ENV DEBIAN_FRONTEND=noninteractive
+ENV GRADIO_SERVER_NAME="0.0.0.0"
+
+# Install necessary packages
+RUN apt-get update && apt-get install -y \
+    python3 \
+    python3-pip \
+    openssh-client \
+    build-essential \
+    git
+
+COPY . /lang-segment-anything
+
+# Install dependencies
+WORKDIR /lang-segment-anything
+RUN pip install -r requirements.txt
+
+EXPOSE 8000
+
+# Entry point
+CMD ["python3", "app.py"]

LICENSE

@@ -0,0 +1,201 @@
+                                 Apache License
+                           Version 2.0, January 2004
+                        http://www.apache.org/licenses/
+
+   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
+
+   1. Definitions.
+
+      "License" shall mean the terms and conditions for use, reproduction,
+      and distribution as defined by Sections 1 through 9 of this document.
+
+      "Licensor" shall mean the copyright owner or entity authorized by
+      the copyright owner that is granting the License.
+
+      "Legal Entity" shall mean the union of the acting entity and all
+      other entities that control, are controlled by, or are under common
+      control with that entity. For the purposes of this definition,
+      "control" means (i) the power, direct or indirect, to cause the
+      direction or management of such entity, whether by contract or
+      otherwise, or (ii) ownership of fifty percent (50%) or more of the
+      outstanding shares, or (iii) beneficial ownership of such entity.
+
+      "You" (or "Your") shall mean an individual or Legal Entity
+      exercising permissions granted by this License.
+
+      "Source" form shall mean the preferred form for making modifications,
+      including but not limited to software source code, documentation
+      source, and configuration files.
+
+      "Object" form shall mean any form resulting from mechanical
+      transformation or translation of a Source form, including but
+      not limited to compiled object code, generated documentation,
+      and conversions to other media types.
+
+      "Work" shall mean the work of authorship, whether in Source or
+      Object form, made available under the License, as indicated by a
+      copyright notice that is included in or attached to the work
+      (an example is provided in the Appendix below).
+
+      "Derivative Works" shall mean any work, whether in Source or Object
+      form, that is based on (or derived from) the Work and for which the
+      editorial revisions, annotations, elaborations, or other modifications
+      represent, as a whole, an original work of authorship. For the purposes
+      of this License, Derivative Works shall not include works that remain
+      separable from, or merely link (or bind by name) to the interfaces of,
+      the Work and Derivative Works thereof.
+
+      "Contribution" shall mean any work of authorship, including
+      the original version of the Work and any modifications or additions
+      to that Work or Derivative Works thereof, that is intentionally
+      submitted to Licensor for inclusion in the Work by the copyright owner
+      or by an individual or Legal Entity authorized to submit on behalf of
+      the copyright owner. For the purposes of this definition, "submitted"
+      means any form of electronic, verbal, or written communication sent
+      to the Licensor or its representatives, including but not limited to
+      communication on electronic mailing lists, source code control systems,
+      and issue tracking systems that are managed by, or on behalf of, the
+      Licensor for the purpose of discussing and improving the Work, but
+      excluding communication that is conspicuously marked or otherwise
+      designated in writing by the copyright owner as "Not a Contribution."
+
+      "Contributor" shall mean Licensor and any individual or Legal Entity
+      on behalf of whom a Contribution has been received by Licensor and
+      subsequently incorporated within the Work.
+
+   2. Grant of Copyright License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      copyright license to reproduce, prepare Derivative Works of,
+      publicly display, publicly perform, sublicense, and distribute the
+      Work and such Derivative Works in Source or Object form.
+
+   3. Grant of Patent License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      (except as stated in this section) patent license to make, have made,
+      use, offer to sell, sell, import, and otherwise transfer the Work,
+      where such license applies only to those patent claims licensable
+      by such Contributor that are necessarily infringed by their
+      Contribution(s) alone or by combination of their Contribution(s)
+      with the Work to which such Contribution(s) was submitted. If You
+      institute patent litigation against any entity (including a
+      cross-claim or counterclaim in a lawsuit) alleging that the Work
+      or a Contribution incorporated within the Work constitutes direct
+      or contributory patent infringement, then any patent licenses
+      granted to You under this License for that Work shall terminate
+      as of the date such litigation is filed.
+
+   4. Redistribution. You may reproduce and distribute copies of the
+      Work or Derivative Works thereof in any medium, with or without
+      modifications, and in Source or Object form, provided that You
+      meet the following conditions:
+
+      (a) You must give any other recipients of the Work or
+          Derivative Works a copy of this License; and
+
+      (b) You must cause any modified files to carry prominent notices
+          stating that You changed the files; and
+
+      (c) You must retain, in the Source form of any Derivative Works
+          that You distribute, all copyright, patent, trademark, and
+          attribution notices from the Source form of the Work,
+          excluding those notices that do not pertain to any part of
+          the Derivative Works; and
+
+      (d) If the Work includes a "NOTICE" text file as part of its
+          distribution, then any Derivative Works that You distribute must
+          include a readable copy of the attribution notices contained
+          within such NOTICE file, excluding those notices that do not
+          pertain to any part of the Derivative Works, in at least one
+          of the following places: within a NOTICE text file distributed
+          as part of the Derivative Works; within the Source form or
+          documentation, if provided along with the Derivative Works; or,
+          within a display generated by the Derivative Works, if and
+          wherever such third-party notices normally appear. The contents
+          of the NOTICE file are for informational purposes only and
+          do not modify the License. You may add Your own attribution
+          notices within Derivative Works that You distribute, alongside
+          or as an addendum to the NOTICE text from the Work, provided
+          that such additional attribution notices cannot be construed
+          as modifying the License.
+
+      You may add Your own copyright statement to Your modifications and
+      may provide additional or different license terms and conditions
+      for use, reproduction, or distribution of Your modifications, or
+      for any such Derivative Works as a whole, provided Your use,
+      reproduction, and distribution of the Work otherwise complies with
+      the conditions stated in this License.
+
+   5. Submission of Contributions. Unless You explicitly state otherwise,
+      any Contribution intentionally submitted for inclusion in the Work
+      by You to the Licensor shall be under the terms and conditions of
+      this License, without any additional terms or conditions.
+      Notwithstanding the above, nothing herein shall supersede or modify
+      the terms of any separate license agreement you may have executed
+      with Licensor regarding such Contributions.
+
+   6. Trademarks. This License does not grant permission to use the trade
+      names, trademarks, service marks, or product names of the Licensor,
+      except as required for reasonable and customary use in describing the
+      origin of the Work and reproducing the content of the NOTICE file.
+
+   7. Disclaimer of Warranty. Unless required by applicable law or
+      agreed to in writing, Licensor provides the Work (and each
+      Contributor provides its Contributions) on an "AS IS" BASIS,
+      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
+      implied, including, without limitation, any warranties or conditions
+      of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
+      PARTICULAR PURPOSE. You are solely responsible for determining the
+      appropriateness of using or redistributing the Work and assume any
+      risks associated with Your exercise of permissions under this License.
+
+   8. Limitation of Liability. In no event and under no legal theory,
+      whether in tort (including negligence), contract, or otherwise,
+      unless required by applicable law (such as deliberate and grossly
+      negligent acts) or agreed to in writing, shall any Contributor be
+      liable to You for damages, including any direct, indirect, special,
+      incidental, or consequential damages of any character arising as a
+      result of this License or out of the use or inability to use the
+      Work (including but not limited to damages for loss of goodwill,
+      work stoppage, computer failure or malfunction, or any and all
+      other commercial damages or losses), even if such Contributor
+      has been advised of the possibility of such damages.
+
+   9. Accepting Warranty or Additional Liability. While redistributing
+      the Work or Derivative Works thereof, You may choose to offer,
+      and charge a fee for, acceptance of support, warranty, indemnity,
+      or other liability obligations and/or rights consistent with this
+      License. However, in accepting such obligations, You may act only
+      on Your own behalf and on Your sole responsibility, not on behalf
+      of any other Contributor, and only if You agree to indemnify,
+      defend, and hold each Contributor harmless for any liability
+      incurred by, or claims asserted against, such Contributor by reason
+      of your accepting any such warranty or additional liability.
+
+   END OF TERMS AND CONDITIONS
+
+   APPENDIX: How to apply the Apache License to your work.
+
+      To apply the Apache License to your work, attach the following
+      boilerplate notice, with the fields enclosed by brackets "[]"
+      replaced with your own identifying information. (Don't include
+      the brackets!)  The text should be enclosed in the appropriate
+      comment syntax for the file format. We also recommend that a
+      file or class name and description of purpose be included on the
+      same "printed page" as the copyright notice for easier
+      identification within third-party archives.
+
+   Copyright [yyyy] [name of copyright owner]
+
+   Licensed under the Apache License, Version 2.0 (the "License");
+   you may not use this file except in compliance with the License.
+   You may obtain a copy of the License at
+
+       http://www.apache.org/licenses/LICENSE-2.0
+
+   Unless required by applicable law or agreed to in writing, software
+   distributed under the License is distributed on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+   See the License for the specific language governing permissions and
+   limitations under the License.

README.md

@@ -1,12 +1,98 @@
----
-title: Lang Segment Anything Gradio
-emoji: 📚
-colorFrom: pink
-colorTo: yellow
-sdk: gradio
-sdk_version: 5.7.0
-app_file: app.py
-pinned: false
----
-
-Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+# Language Segment-Anything
+
+Language Segment-Anything is an open-source project that combines the power of instance segmentation and text prompts to generate masks for specific objects in images. Built on the recently released Meta model, Segment Anything Model 2, and the GroundingDINO detection model, it's an easy-to-use and effective tool for object detection and image segmentation.
+
+![person.png](/assets/outputs/person.png)
+
+## Features
+
+- Zero-shot text-to-bbox approach for object detection.
+- GroundingDINO detection model integration.
+- SAM 2.1
+- Batch inference support.
+- Easy endpoint deployment using the Lightning AI litserve platform.
+- Customizable text prompts for precise object segmentation.
+
+## Getting Started
+
+### Prerequisites
+
+- Python 3.11 or higher
+
+### Installation
+
+#### Installing PyTorch Dependencies
+
+Before installing `lang-sam`, please install PyTorch using the following command:
+
+```bash
+
+pip install torch==2.4.1 torchvision==0.19.1 --extra-index-url https://download.pytorch.org/whl/cu124
+
+```
+
+```bash
+
+pip install -U git+https://github.com/luca-medeiros/lang-segment-anything.git
+
+```
+
+Or
+Clone the repository and install the required packages:
+
+```bash
+
+git clone https://github.com/luca-medeiros/lang-segment-anything && cd lang-segment-anything
+pip install -e .
+
+```
+
+#### Docker Installation
+
+Build and run the image.
+
+```bash
+
+git clone https://github.com/luca-medeiros/lang-segment-anything && cd lang-segment-anything
+docker build --tag lang-segment-anything:latest .
+docker run --gpus all -p 8000:8000 lang-segment-anything:latest
+
+```
+
+### Usage
+
+To run the gradio APP:
+
+`python app.py`
+And open `http://0.0.0.0:8000/gradio`
+
+Use as a library:
+
+```python
+from PIL import Image
+from lang_sam import LangSAM
+
+model = LangSAM()
+image_pil = Image.open("./assets/car.jpeg").convert("RGB")
+text_prompt = "wheel."
+results = model.predict([image_pil], [text_prompt])
+```
+
+## Examples
+
+![car.png](/assets/outputs/car.png)
+
+![fruits.png](/assets/outputs/fruits.png)
+
+## Acknowledgments
+
+This project is based on/used the following repositories:
+
+- [GroundingDINO](https://github.com/IDEA-Research/GroundingDINO)
+- [Segment-Anything](https://github.com/facebookresearch/segment-anything-2)
+- [LitServe](https://github.com/Lightning-AI/LitServe/)
+- [Supervision](https://github.com/roboflow/supervision)
+
+## License
+
+This project is licensed under the Apache 2.0 License

app.py

@@ -1,7 +1,98 @@
+import os
+from io import BytesIO
+
 import gradio as gr
+import requests
+from PIL import Image
+
+from lang_sam import SAM_MODELS
+from lang_sam.server import PORT, server
+
+
+def inference(sam_type, box_threshold, text_threshold, image, text_prompt):
+    """Gradio function that makes a request to the /predict LitServe endpoint."""
+    url = f"http://localhost:{PORT}/predict"  # Adjust port if needed
+
+    # Prepare the multipart form data
+    with open(image, "rb") as img_file:
+        files = {
+            "image": img_file,
+        }
+        data = {
+            "sam_type": sam_type,
+            "box_threshold": str(box_threshold),
+            "text_threshold": str(text_threshold),
+            "text_prompt": text_prompt,
+        }
+
+        try:
+            response = requests.post(url, files=files, data=data)
+        except Exception as e:
+            print(f"Request failed: {e}")
+            return None
+
+    if response.status_code == 200:
+        try:
+            output_image = Image.open(BytesIO(response.content)).convert("RGB")
+            return output_image
+        except Exception as e:
+            print(f"Failed to process response image: {e}")
+            return None
+    else:
+        print(f"Request failed with status code {response.status_code}: {response.text}")
+        return None
+
+
+with gr.Blocks(title="lang-sam") as blocks:
+    with gr.Row():
+        sam_model_choices = gr.Dropdown(choices=list(SAM_MODELS.keys()), label="SAM Model", value="sam2.1_hiera_small")
+        box_threshold = gr.Slider(minimum=0.0, maximum=1.0, value=0.3, label="Box Threshold")
+        text_threshold = gr.Slider(minimum=0.0, maximum=1.0, value=0.25, label="Text Threshold")
+    with gr.Row():
+        image_input = gr.Image(type="filepath", label="Input Image")
+        output_image = gr.Image(type="pil", label="Output Image")
+    text_prompt = gr.Textbox(lines=1, label="Text Prompt")
+
+    submit_btn = gr.Button("Run Prediction")
+
+    submit_btn.click(
+        fn=inference,
+        inputs=[sam_model_choices, box_threshold, text_threshold, image_input, text_prompt],
+        outputs=output_image,
+    )
+
+    examples = [
+        [
+            "sam2.1_hiera_small",
+            0.32,
+            0.25,
+            os.path.join(os.path.dirname(__file__), "assets", "fruits.jpg"),
+            "kiwi. watermelon. blueberry.",
+        ],
+        [
+            "sam2.1_hiera_small",
+            0.3,
+            0.25,
+            os.path.join(os.path.dirname(__file__), "assets", "car.jpeg"),
+            "wheel.",
+        ],
+        [
+            "sam2.1_hiera_small",
+            0.3,
+            0.25,
+            os.path.join(os.path.dirname(__file__), "assets", "food.jpg"),
+            "food.",
+        ],
+    ]
+
+    gr.Examples(
+        examples=examples,
+        inputs=[sam_model_choices, box_threshold, text_threshold, image_input, text_prompt],
+        outputs=output_image,
+    )
 
-def greet(name):
-    return "Hello " + name + "!!"
+server.app = gr.mount_gradio_app(server.app, blocks, path="/gradio")
 
-demo = gr.Interface(fn=greet, inputs="text", outputs="text")
-demo.launch()
+if __name__ == "__main__":
+    print(f"Starting LitServe and Gradio server on port {PORT}...")
+    server.run(port=PORT)

lang_sam/__init__.py

@@ -0,0 +1,2 @@
+from lang_sam.lang_sam import LangSAM
+from lang_sam.models.sam import SAM_MODELS

lang_sam/lang_sam.py

@@ -0,0 +1,82 @@
+import numpy as np
+from PIL import Image
+
+from lang_sam.models.gdino import GDINO
+from lang_sam.models.sam import SAM
+
+
+class LangSAM:
+    def __init__(self, sam_type="sam2.1_hiera_small", ckpt_path: str | None = None):
+        self.sam_type = sam_type
+        self.sam = SAM()
+        self.sam.build_model(sam_type, ckpt_path)
+        self.gdino = GDINO()
+        self.gdino.build_model()
+
+    def predict(
+        self,
+        images_pil: list[Image.Image],
+        texts_prompt: list[str],
+        box_threshold: float = 0.3,
+        text_threshold: float = 0.25,
+    ):
+        """Predicts masks for given images and text prompts using GDINO and SAM models.
+
+        Parameters:
+            images_pil (list[Image.Image]): List of input images.
+            texts_prompt (list[str]): List of text prompts corresponding to the images.
+            box_threshold (float): Threshold for box predictions.
+            text_threshold (float): Threshold for text predictions.
+
+        Returns:
+            list[dict]: List of results containing masks and other outputs for each image.
+            Output format:
+            [{
+                "boxes": np.ndarray,
+                "scores": np.ndarray,
+                "masks": np.ndarray,
+                "mask_scores": np.ndarray,
+            }, ...]
+        """
+
+        gdino_results = self.gdino.predict(images_pil, texts_prompt, box_threshold, text_threshold)
+        all_results = []
+        sam_images = []
+        sam_boxes = []
+        sam_indices = []
+        for idx, result in enumerate(gdino_results):
+            processed_result = {
+                **result,
+                "masks": [],
+                "mask_scores": [],
+            }
+
+            if result["labels"]:
+                processed_result["boxes"] = result["boxes"].cpu().numpy()
+                processed_result["scores"] = result["scores"].cpu().numpy()
+                sam_images.append(np.asarray(images_pil[idx]))
+                sam_boxes.append(processed_result["boxes"])
+                sam_indices.append(idx)
+
+            all_results.append(processed_result)
+        if sam_images:
+            print(f"Predicting {len(sam_boxes)} masks")
+            masks, mask_scores, _ = self.sam.predict_batch(sam_images, xyxy=sam_boxes)
+            for idx, mask, score in zip(sam_indices, masks, mask_scores):
+                all_results[idx].update(
+                    {
+                        "masks": mask,
+                        "mask_scores": score,
+                    }
+                )
+            print(f"Predicted {len(all_results)} masks")
+        return all_results
+
+
+if __name__ == "__main__":
+    model = LangSAM()
+    out = model.predict(
+        [Image.open("./assets/food.jpg"), Image.open("./assets/car.jpeg")],
+        ["food", "car"],
+    )
+    print(out)

lang_sam/models/gdino.py

@@ -0,0 +1,61 @@
+import torch
+from PIL import Image
+from transformers import AutoModelForZeroShotObjectDetection, AutoProcessor
+
+from lang_sam.models.utils import get_device_type
+
+device_type = get_device_type()
+DEVICE = torch.device(device_type)
+
+if torch.cuda.is_available():
+    torch.autocast(device_type="cuda", dtype=torch.bfloat16).__enter__()
+    if torch.cuda.get_device_properties(0).major >= 8:
+        torch.backends.cuda.matmul.allow_tf32 = True
+        torch.backends.cudnn.allow_tf32 = True
+
+
+class GDINO:
+    def __init__(self):
+        self.build_model()
+
+    def build_model(self, ckpt_path: str | None = None):
+        model_id = "IDEA-Research/grounding-dino-base"
+        self.processor = AutoProcessor.from_pretrained(model_id)
+        self.model = AutoModelForZeroShotObjectDetection.from_pretrained(model_id).to(
+            DEVICE
+        )
+
+    def predict(
+        self,
+        pil_images: list[Image.Image],
+        text_prompt: list[str],
+        box_threshold: float,
+        text_threshold: float,
+    ) -> list[dict]:
+        for i, prompt in enumerate(text_prompt):
+            if prompt[-1] != ".":
+                text_prompt[i] += "."
+        inputs = self.processor(images=pil_images, text=text_prompt, return_tensors="pt").to(DEVICE)
+        with torch.no_grad():
+            outputs = self.model(**inputs)
+
+        results = self.processor.post_process_grounded_object_detection(
+            outputs,
+            inputs.input_ids,
+            box_threshold=box_threshold,
+            text_threshold=text_threshold,
+            target_sizes=[k.size[::-1] for k in pil_images],
+        )
+        return results
+
+
+if __name__ == "__main__":
+    gdino = GDINO()
+    gdino.build_model()
+    out = gdino.predict(
+        [Image.open("./assets/car.jpeg"), Image.open("./assets/car.jpeg")],
+        ["wheel", "wheel"],
+        0.3,
+        0.25,
+    )
+    print(out)

lang_sam/models/sam.py

@@ -0,0 +1,103 @@
+import numpy as np
+import torch
+from hydra import compose
+from hydra.utils import instantiate
+from omegaconf import OmegaConf
+from sam2.automatic_mask_generator import SAM2AutomaticMaskGenerator
+from sam2.sam2_image_predictor import SAM2ImagePredictor
+
+from lang_sam.models.utils import get_device_type
+
+DEVICE = torch.device(get_device_type())
+
+if torch.cuda.is_available():
+    torch.autocast(device_type="cuda", dtype=torch.bfloat16).__enter__()
+    if torch.cuda.get_device_properties(0).major >= 8:
+        torch.backends.cuda.matmul.allow_tf32 = True
+        torch.backends.cudnn.allow_tf32 = True
+
+
+SAM_MODELS = {
+    "sam2.1_hiera_tiny": {
+        "url": "https://dl.fbaipublicfiles.com/segment_anything_2/092824/sam2.1_hiera_tiny.pt",
+        "config": "configs/sam2.1/sam2.1_hiera_t.yaml",
+    },
+    "sam2.1_hiera_small": {
+        "url": "https://dl.fbaipublicfiles.com/segment_anything_2/092824/sam2.1_hiera_small.pt",
+        "config": "configs/sam2.1/sam2.1_hiera_s.yaml",
+    },
+    "sam2.1_hiera_base_plus": {
+        "url": "https://dl.fbaipublicfiles.com/segment_anything_2/092824/sam2.1_hiera_base_plus.pt",
+        "config": "configs/sam2.1/sam2.1_hiera_b+.yaml",
+    },
+    "sam2.1_hiera_large": {
+        "url": "https://dl.fbaipublicfiles.com/segment_anything_2/092824/sam2.1_hiera_large.pt",
+        "config": "configs/sam2.1/sam2.1_hiera_l.yaml",
+    },
+}
+
+
+class SAM:
+    def build_model(self, sam_type: str, ckpt_path: str | None = None):
+        self.sam_type = sam_type
+        self.ckpt_path = ckpt_path
+        cfg = compose(config_name=SAM_MODELS[self.sam_type]["config"], overrides=[])
+        OmegaConf.resolve(cfg)
+        self.model = instantiate(cfg.model, _recursive_=True)
+        self._load_checkpoint(self.model)
+        self.model = self.model.to(DEVICE)
+        self.model.eval()
+        self.mask_generator = SAM2AutomaticMaskGenerator(self.model)
+        self.predictor = SAM2ImagePredictor(self.model)
+
+    def _load_checkpoint(self, model: torch.nn.Module):
+        if self.ckpt_path is None:
+            checkpoint_url = SAM_MODELS[self.sam_type]["url"]
+            state_dict = torch.hub.load_state_dict_from_url(checkpoint_url, map_location="cpu")["model"]
+        else:
+            state_dict = torch.load(self.ckpt_path, map_location="cpu", weights_only=True)
+        try:
+            model.load_state_dict(state_dict, strict=True)
+        except Exception as e:
+            raise ValueError(f"Problem loading SAM please make sure you have the right model type: {self.sam_type} \
+                and a working checkpoint: {checkpoint_url}. Recommend deleting the checkpoint and \
+                re-downloading it. Error: {e}")
+
+    def generate(self, image_rgb: np.ndarray) -> list[dict]:
+        """
+        Output format
+        SAM2AutomaticMaskGenerator returns a list of masks, where each mask is a dict containing various information
+        about the mask:
+
+        segmentation - [np.ndarray] - the mask with (W, H) shape, and bool type
+        area - [int] - the area of the mask in pixels
+        bbox - [List[int]] - the boundary box of the mask in xywh format
+        predicted_iou - [float] - the model's own prediction for the quality of the mask
+        point_coords - [List[List[float]]] - the sampled input point that generated this mask
+        stability_score - [float] - an additional measure of mask quality
+        crop_box - List[int] - the crop of the image used to generate this mask in xywh format
+        """
+
+        sam2_result = self.mask_generator.generate(image_rgb)
+        return sam2_result
+
+    def predict(self, image_rgb: np.ndarray, xyxy: np.ndarray) -> tuple[np.ndarray, np.ndarray, np.ndarray]:
+        self.predictor.set_image(image_rgb)
+        masks, scores, logits = self.predictor.predict(box=xyxy, multimask_output=False)
+        if len(masks.shape) > 3:
+            masks = np.squeeze(masks, axis=1)
+        return masks, scores, logits
+
+    def predict_batch(
+        self,
+        images_rgb: list[np.ndarray],
+        xyxy: list[np.ndarray],
+    ) -> tuple[list[np.ndarray], list[np.ndarray], list[np.ndarray]]:
+        self.predictor.set_image_batch(images_rgb)
+
+        masks, scores, logits = self.predictor.predict_batch(box_batch=xyxy, multimask_output=False)
+
+        masks = [np.squeeze(mask, axis=1) if len(mask.shape) > 3 else mask for mask in masks]
+        scores = [np.squeeze(score) for score in scores]
+        logits = [np.squeeze(logit, axis=1) if len(logit.shape) > 3 else logit for logit in logits]
+        return masks, scores, logits

lang_sam/models/utils.py

@@ -0,0 +1,13 @@
+import logging
+
+import torch
+
+
+def get_device_type() -> str:
+    if torch.backends.mps.is_available():
+        return "mps"
+    elif torch.cuda.is_available():
+        return "cuda"
+    else:
+        logging.warning("No GPU found, using CPU instead")
+        return "cpu"

lang_sam/server.py

@@ -0,0 +1,121 @@
+from io import BytesIO
+
+import litserve as ls
+import numpy as np
+from fastapi import Response, UploadFile
+from PIL import Image
+
+from lang_sam import LangSAM
+from lang_sam.utils import draw_image
+
+PORT = 8000
+
+
+class LangSAMAPI(ls.LitAPI):
+    def setup(self, device: str) -> None:
+        """Initialize or load the LangSAM model."""
+        self.model = LangSAM(sam_type="sam2.1_hiera_small")
+        print("LangSAM model initialized.")
+
+    def decode_request(self, request) -> dict:
+        """Decode the incoming request to extract parameters and image bytes.
+
+        Assumes the request is sent as multipart/form-data with fields:
+        - sam_type: str
+        - box_threshold: float
+        - text_threshold: float
+        - text_prompt: str
+        - image: UploadFile
+        """
+        # Extract form data
+        sam_type = request.get("sam_type")
+        box_threshold = float(request.get("box_threshold", 0.3))
+        text_threshold = float(request.get("text_threshold", 0.25))
+        text_prompt = request.get("text_prompt", "")
+
+        # Extract image file
+        image_file: UploadFile = request.get("image")
+        if image_file is None:
+            raise ValueError("No image file provided in the request.")
+
+        image_bytes = image_file.file.read()
+
+        return {
+            "sam_type": sam_type,
+            "box_threshold": box_threshold,
+            "text_threshold": text_threshold,
+            "image_bytes": image_bytes,
+            "text_prompt": text_prompt,
+        }
+
+    def predict(self, inputs: dict) -> dict:
+        """Perform prediction using the LangSAM model.
+
+        Yields:
+            dict: Contains the processed output image.
+        """
+        print("Starting prediction with parameters:")
+        print(
+            f"sam_type: {inputs['sam_type']}, \
+                box_threshold: {inputs['box_threshold']}, \
+                text_threshold: {inputs['text_threshold']}, \
+                text_prompt: {inputs['text_prompt']}"
+        )
+
+        if inputs["sam_type"] != self.model.sam_type:
+            print(f"Updating SAM model type to {inputs['sam_type']}")
+            self.model.sam.build_model(inputs["sam_type"])
+
+        try:
+            image_pil = Image.open(BytesIO(inputs["image_bytes"])).convert("RGB")
+        except Exception as e:
+            raise ValueError(f"Invalid image data: {e}")
+
+        results = self.model.predict(
+            images_pil=[image_pil],
+            texts_prompt=[inputs["text_prompt"]],
+            box_threshold=inputs["box_threshold"],
+            text_threshold=inputs["text_threshold"],
+        )
+        results = results[0]
+
+        if not len(results["masks"]):
+            print("No masks detected. Returning original image.")
+            return {"output_image": image_pil}
+
+        # Draw results on the image
+        image_array = np.asarray(image_pil)
+        output_image = draw_image(
+            image_array,
+            results["masks"],
+            results["boxes"],
+            results["scores"],
+            results["labels"],
+        )
+        output_image = Image.fromarray(np.uint8(output_image)).convert("RGB")
+
+        return {"output_image": output_image}
+
+    def encode_response(self, output: dict) -> Response:
+        """Encode the prediction result into an HTTP response.
+
+        Returns:
+            Response: Contains the processed image in PNG format.
+        """
+        try:
+            image = output["output_image"]
+            buffer = BytesIO()
+            image.save(buffer, format="PNG")
+            buffer.seek(0)
+            return Response(content=buffer.getvalue(), media_type="image/png")
+        except StopIteration:
+            raise ValueError("No output generated by the prediction.")
+
+
+lit_api = LangSAMAPI()
+server = ls.LitServer(lit_api)
+
+
+if __name__ == "__main__":
+    print(f"Starting LitServe and Gradio server on port {PORT}...")
+    server.run(port=PORT)

lang_sam/utils.py

@@ -0,0 +1,99 @@
+import cv2
+import numpy as np
+import supervision as sv
+from PIL import Image
+
+MIN_AREA = 100
+
+
+def load_image(image_path: str):
+    return Image.open(image_path).convert("RGB")
+
+
+def draw_image(image_rgb, masks, xyxy, probs, labels):
+    box_annotator = sv.BoxCornerAnnotator()
+    label_annotator = sv.LabelAnnotator()
+    mask_annotator = sv.MaskAnnotator()
+    # Create class_id for each unique label
+    unique_labels = list(set(labels))
+    class_id_map = {label: idx for idx, label in enumerate(unique_labels)}
+    class_id = [class_id_map[label] for label in labels]
+
+    # Add class_id to the Detections object
+    detections = sv.Detections(
+        xyxy=xyxy,
+        mask=masks.astype(bool),
+        confidence=probs,
+        class_id=np.array(class_id),
+    )
+    annotated_image = box_annotator.annotate(scene=image_rgb.copy(), detections=detections)
+    annotated_image = label_annotator.annotate(scene=annotated_image, detections=detections, labels=labels)
+    annotated_image = mask_annotator.annotate(scene=annotated_image, detections=detections)
+    return annotated_image
+
+
+def get_contours(mask):
+    if len(mask.shape) > 2:
+        mask = np.squeeze(mask, 0)
+    mask = mask.astype(np.uint8)
+    mask *= 255
+    contours, _ = cv2.findContours(mask, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
+    effContours = []
+    for c in contours:
+        area = cv2.contourArea(c)
+        if area > MIN_AREA:
+            effContours.append(c)
+    return effContours
+
+
+def contour_to_points(contour):
+    pointsNum = len(contour)
+    contour = contour.reshape(pointsNum, -1).astype(np.float32)
+    points = [point.tolist() for point in contour]
+    return points
+
+
+def generate_labelme_json(binary_masks, labels, image_size, image_path=None):
+    """Generate a LabelMe format JSON file from binary mask tensor.
+
+    Args:
+        binary_masks: Binary mask tensor of shape [N, H, W].
+        labels: List of labels for each mask.
+        image_size: Tuple of (height, width) for the image size.
+        image_path: Path to the image file (optional).
+
+    Returns:
+        A dictionary representing the LabelMe JSON file.
+    """
+    num_masks = binary_masks.shape[0]
+    binary_masks = binary_masks.numpy()
+
+    json_dict = {
+        "version": "4.5.6",
+        "imageHeight": image_size[0],
+        "imageWidth": image_size[1],
+        "imagePath": image_path,
+        "flags": {},
+        "shapes": [],
+        "imageData": None,
+    }
+
+    # Loop through the masks and add them to the JSON dictionary
+    for i in range(num_masks):
+        mask = binary_masks[i]
+        label = labels[i]
+        effContours = get_contours(mask)
+
+        for effContour in effContours:
+            points = contour_to_points(effContour)
+            shape_dict = {
+                "label": label,
+                "line_color": None,
+                "fill_color": None,
+                "points": points,
+                "shape_type": "polygon",
+            }
+
+            json_dict["shapes"].append(shape_dict)
+
+    return json_dict

pyproject.toml

@@ -0,0 +1,59 @@
+[build-system]
+requires = ["setuptools", "wheel"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "lang-sam"
+version = "0.2.1"
+description = "Language segment-anything"
+readme = "README.md"
+authors = [
+  { name = "Luca Medeiros", email = "[email protected]" },
+]
+requires-python = ">=3.11"
+dynamic = ["dependencies"]
+
+[tool.setuptools.dynamic]
+dependencies = {file = ["requirements.txt"]}
+
+[tool.setuptools.packages.find]
+where = ["."]
+include = ["lang_sam", "lang_sam.*"]
+
+[tool.ruff]
+target-version = "py311"
+line-length = 120
+fix = true
+select = [
+    # https://github.com/charliermarsh/ruff#pyflakes-f
+    "F", # Pyflakes
+    # https://github.com/charliermarsh/ruff#pycodestyle-e-w
+    "E", # pycodestyle
+    "W", # Warning
+    # https://github.com/charliermarsh/ruff#flake8-comprehensions-c4
+    # https://github.com/charliermarsh/ruff#mccabe-c90
+    "C", # Complexity (mccabe+) & comprehensions
+    # https://github.com/charliermarsh/ruff#pyupgrade-up
+    "UP", # pyupgrade
+    # https://github.com/charliermarsh/ruff#isort-i
+    "I", # isort
+]
+ignore = [
+    # https://github.com/charliermarsh/ruff#pycodestyle-e-w
+    "E402", # module level import not at top of file
+    # https://github.com/charliermarsh/ruff#pyupgrade-up
+    "UP006", # use-pep585-annotation
+    "UP007", # use-pep604-annotation
+    "E741", # Ambiguous variable name
+]
+[tool.ruff.per-file-ignores]
+"__init__.py" = [
+    "F401",  # unused import
+    "F403",  # star imports
+    ]
+
+[tool.ruff.mccabe]
+max-complexity = 24
+
+[tool.ruff.pydocstyle]
+convention = "numpy"

requirements.txt

@@ -0,0 +1,10 @@
+gradio==5.0.2
+litserve==0.2.3
+opencv-python-headless==4.10.0.84
+pydantic==2.9.2
+sam-2 @ git+https://github.com/facebookresearch/segment-anything-2@c2ec8e14a185632b0a5d8b161928ceb50197eddc
+supervision==0.23.0 ; python_full_version > '3.10'
+transformers==4.44.2
+uvloop==0.20.0
+torch==2.4.1
+torchvision==0.19.1