vision-agent 0.2.235__py3-none-any.whl → 0.2.237__py3-none-any.whl

vision_agent/utils/__init__.py

@@ -6,4 +6,3 @@ from .execute import (
     Logs,
     Result,
 )
-from .sim import AzureSim, OllamaSim, Sim, load_sim, merge_sim

vision_agent/{agent/agent_utils.py → utils/agent.py}

@@ -1,3 +1,4 @@
+import base64
 import copy
 import json
 import logging
@@ -14,8 +15,7 @@ from rich.syntax import Syntax
 from rich.table import Table
 
 import vision_agent.tools as T
-from vision_agent.agent.types import AgentMessage, PlanContext
-from vision_agent.lmm.types import Message
+from vision_agent.models import AgentMessage, Message, PlanContext
 from vision_agent.utils.execute import CodeInterpreter, Execution
 from vision_agent.utils.image_utils import b64_to_pil, convert_to_b64
 
@@ -185,6 +185,7 @@ class DefaultImports:
         "import os",
         "import numpy as np",
         "from vision_agent.tools import *",
+        "from vision_agent.tools.planner_tools import judge_od_results",
         "from typing import *",
         "from pillow_heif import register_heif_opener",
         "register_heif_opener()",
@@ -248,6 +249,14 @@ def add_media_to_chat(
                     ) as temp_file:
                         media_pil.save(temp_file, format="PNG")
                         media = str(temp_file.name)
+                elif isinstance(media, str) and media.startswith("data:video/"):
+                    ext = media.split(";")[0].split("/")[-1]
+                    with tempfile.NamedTemporaryFile(
+                        mode="wb", suffix=f".{ext}", delete=False
+                    ) as temp_file:
+                        media_bytes = base64.b64decode(media.split(",")[1])
+                        temp_file.write(media_bytes)
+                        media = str(temp_file.name)
                 if code_interpreter is not None:
                     media = str(code_interpreter.upload_file(media))
                 media_list_i.append(media)

vision_agent/utils/image_utils.py

@@ -183,7 +183,9 @@ def encode_image_bytes(image: bytes, resize: Optional[int] = None) -> str:
     return encoded_image
 
 
-def encode_media(media: Union[str, Path], resize: Optional[int] = None) -> str:
+def encode_media(
+    media: Union[str, Path, np.ndarray, ImageType], resize: Optional[int] = None
+) -> str:
     if isinstance(media, str) and media.startswith(("http", "https")):
         # for mp4 video url, we assume there is a same url but ends with png
         # vision-agent-ui will upload this png when uploading the video
@@ -191,16 +193,25 @@ def encode_media(media: Union[str, Path], resize: Optional[int] = None) -> str:
             return media[:-4] + ".png"
         return media
 
-    # if media is in base64 ensure it's the correct resize
-    if isinstance(media, str) and media.startswith("data:image/"):
-        image_pil = b64_to_pil(media)
+    def resize_to_b64(image: ImageType, resize: Optional[int] = None) -> str:
         if resize is not None:
-            if image_pil.size[0] > resize or image_pil.size[1] > resize:
-                image_pil.thumbnail((resize, resize))
+            image.thumbnail((resize, resize))
         buffer = io.BytesIO()
-        image_pil.save(buffer, format="PNG")
+        image.save(buffer, format="PNG")
         return base64.b64encode(buffer.getvalue()).decode("utf-8")
 
+    # if media is in base64, numpy array or PIL Image ensure it's the correct resize
+    if isinstance(media, str) and media.startswith("data:image/"):
+        image_pil = b64_to_pil(media)
+        return resize_to_b64(image_pil, resize=resize)
+
+    if isinstance(media, np.ndarray):
+        image_pil = Image.fromarray(media).convert("RGB")
+        return resize_to_b64(image_pil, resize=resize)
+
+    if isinstance(media, ImageType):
+        return resize_to_b64(media, resize=resize)
+
     extension = "png"
     extension = Path(media).suffix
     if extension.lower() not in {

vision_agent/{tools/tool_utils.py → utils/tools.py}

@@ -1,18 +1,15 @@
-import inspect
 import logging
 import os
 from base64 import b64encode
-from typing import Any, Callable, Dict, List, MutableMapping, Optional, Tuple
+from typing import Any, Dict, List, MutableMapping, Optional, Tuple
 
 import numpy as np
-import pandas as pd
 from IPython.display import display
 from pydantic import BaseModel
 from requests import Session
 from requests.adapters import HTTPAdapter
 from urllib3.util.retry import Retry
 
-from vision_agent.tools.tools_types import BoundingBoxes
 from vision_agent.utils.exceptions import RemoteToolCallFailed
 from vision_agent.utils.execute import Error, MimeType
 from vision_agent.utils.image_utils import normalize_bbox
@@ -121,89 +118,6 @@ def _create_requests_session(
     return session
 
 
-def get_tool_documentation(funcs: List[Callable[..., Any]]) -> str:
-    docstrings = ""
-    for func in funcs:
-        docstrings += f"{func.__name__}{inspect.signature(func)}:\n{func.__doc__}\n\n"
-
-    return docstrings
-
-
-def get_tool_descriptions(funcs: List[Callable[..., Any]]) -> str:
-    descriptions = ""
-    for func in funcs:
-        description = func.__doc__
-        if description is None:
-            description = ""
-
-        if "Parameters:" in description:
-            description = (
-                description[: description.find("Parameters:")]
-                .replace("\n", " ")
-                .strip()
-            )
-
-        description = " ".join(description.split())
-        descriptions += f"- {func.__name__}{inspect.signature(func)}: {description}\n"
-    return descriptions
-
-
-def get_tool_descriptions_by_names(
-    tool_name: Optional[List[str]],
-    funcs: List[Callable[..., Any]],
-    util_funcs: List[
-        Callable[..., Any]
-    ],  # util_funcs will always be added to the list of functions
-) -> str:
-    if tool_name is None:
-        return get_tool_descriptions(funcs + util_funcs)
-
-    invalid_names = [
-        name for name in tool_name if name not in {func.__name__ for func in funcs}
-    ]
-
-    if invalid_names:
-        raise ValueError(f"Invalid customized tool names: {', '.join(invalid_names)}")
-
-    filtered_funcs = (
-        funcs
-        if not tool_name
-        else [func for func in funcs if func.__name__ in tool_name]
-    )
-    return get_tool_descriptions(filtered_funcs + util_funcs)
-
-
-def get_tools_df(funcs: List[Callable[..., Any]]) -> pd.DataFrame:
-    data: Dict[str, List[str]] = {"desc": [], "doc": [], "name": []}
-
-    for func in funcs:
-        desc = func.__doc__
-        if desc is None:
-            desc = ""
-        desc = desc[: desc.find("Parameters:")].replace("\n", " ").strip()
-        desc = " ".join(desc.split())
-
-        doc = f"{func.__name__}{inspect.signature(func)}:\n{func.__doc__}"
-        data["desc"].append(desc)
-        data["doc"].append(doc)
-        data["name"].append(func.__name__)
-
-    return pd.DataFrame(data)  # type: ignore
-
-
-def get_tools_info(funcs: List[Callable[..., Any]]) -> Dict[str, str]:
-    data: Dict[str, str] = {}
-
-    for func in funcs:
-        desc = func.__doc__
-        if desc is None:
-            desc = ""
-
-        data[func.__name__] = f"{func.__name__}{inspect.signature(func)}:\n{desc}"
-
-    return data
-
-
 def _call_post(
     url: str,
     payload: dict[str, Any],
@@ -259,12 +173,6 @@ def _call_post(
             display({MimeType.APPLICATION_JSON: trace}, raw=True)
 
 
-def filter_bboxes_by_threshold(
-    bboxes: BoundingBoxes, threshold: float
-) -> BoundingBoxes:
-    return list(filter(lambda bbox: bbox.score >= threshold, bboxes))
-
-
 def add_bboxes_from_masks(
     all_preds: List[List[Dict[str, Any]]],
 ) -> List[List[Dict[str, Any]]]:

vision_agent/utils/tools_doc.py

@@ -0,0 +1,87 @@
+import inspect
+from typing import Any, Callable, Dict, List, Optional
+
+import pandas as pd
+
+
+def get_tool_documentation(funcs: List[Callable[..., Any]]) -> str:
+    docstrings = ""
+    for func in funcs:
+        docstrings += f"{func.__name__}{inspect.signature(func)}:\n{func.__doc__}\n\n"
+
+    return docstrings
+
+
+def get_tool_descriptions(funcs: List[Callable[..., Any]]) -> str:
+    descriptions = ""
+    for func in funcs:
+        description = func.__doc__
+        if description is None:
+            description = ""
+
+        if "Parameters:" in description:
+            description = (
+                description[: description.find("Parameters:")]
+                .replace("\n", " ")
+                .strip()
+            )
+
+        description = " ".join(description.split())
+        descriptions += f"- {func.__name__}{inspect.signature(func)}: {description}\n"
+    return descriptions
+
+
+def get_tool_descriptions_by_names(
+    tool_name: Optional[List[str]],
+    funcs: List[Callable[..., Any]],
+    util_funcs: List[
+        Callable[..., Any]
+    ],  # util_funcs will always be added to the list of functions
+) -> str:
+    if tool_name is None:
+        return get_tool_descriptions(funcs + util_funcs)
+
+    invalid_names = [
+        name for name in tool_name if name not in {func.__name__ for func in funcs}
+    ]
+
+    if invalid_names:
+        raise ValueError(f"Invalid customized tool names: {', '.join(invalid_names)}")
+
+    filtered_funcs = (
+        funcs
+        if not tool_name
+        else [func for func in funcs if func.__name__ in tool_name]
+    )
+    return get_tool_descriptions(filtered_funcs + util_funcs)
+
+
+def get_tools_df(funcs: List[Callable[..., Any]]) -> pd.DataFrame:
+    data: Dict[str, List[str]] = {"desc": [], "doc": [], "name": []}
+
+    for func in funcs:
+        desc = func.__doc__
+        if desc is None:
+            desc = ""
+        desc = desc[: desc.find("Parameters:")].replace("\n", " ").strip()
+        desc = " ".join(desc.split())
+
+        doc = f"{func.__name__}{inspect.signature(func)}:\n{func.__doc__}"
+        data["desc"].append(desc)
+        data["doc"].append(doc)
+        data["name"].append(func.__name__)
+
+    return pd.DataFrame(data)  # type: ignore
+
+
+def get_tools_info(funcs: List[Callable[..., Any]]) -> Dict[str, str]:
+    data: Dict[str, str] = {}
+
+    for func in funcs:
+        desc = func.__doc__
+        if desc is None:
+            desc = ""
+
+        data[func.__name__] = f"{func.__name__}{inspect.signature(func)}:\n{desc}"
+
+    return data

vision_agent/utils/video.py

@@ -106,6 +106,20 @@ def frames_to_bytes(
     return buffer_bytes
 
 
+def rescale(frame: np.ndarray, max_size: Tuple[int, int]) -> np.ndarray:
+    h, w = frame.shape[:2]
+    new_h, new_w = h, w
+    if new_h > max_size[0]:
+        new_h = max_size[0]
+        new_w = int(w * new_h / h)
+    if new_w > max_size[1]:
+        new_w = max_size[1]
+        new_h = int(h * new_w / w)
+    if h != new_h or w != new_w:
+        frame = cv2.resize(frame, (new_w, new_h))
+    return frame
+
+
 # WARNING: This cache is a little dangerous because if the underlying video
 # contents change but the filename remains the same it will return the old file contents.
 # For vision agent it's unlikely to change the file contents while keeping the
@@ -158,6 +172,7 @@ def extract_frames_from_video(
         # causes the last frame to be skipped
         elapsed_time = round(elapsed_time, 8)
         if elapsed_time >= targ_frame_time:
+            frame = rescale(frame, (1024, 1024))
             frames.append((cv2.cvtColor(frame, cv2.COLOR_BGR2RGB), i / orig_fps))
             elapsed_time -= targ_frame_time
 

vision_agent/utils/video_tracking.py

@@ -5,11 +5,8 @@ from typing import Any, Callable, Dict, List, Optional, Tuple
 import numpy as np
 from scipy.optimize import linear_sum_assignment  # type: ignore
 
-from vision_agent.tools.tool_utils import (
-    add_bboxes_from_masks,
-    send_task_inference_request,
-)
 from vision_agent.utils.image_utils import denormalize_bbox, rle_decode_array
+from vision_agent.utils.tools import add_bboxes_from_masks, send_task_inference_request
 from vision_agent.utils.video import frames_to_bytes
 
 
@@ -115,6 +112,40 @@ def process_segment(
         metadata=metadata,
     )
 
+    segment_detections = join_scores(transformed_detections, segment_detections)
+    return segment_detections
+
+
+def join_scores(
+    transformed_detections: List[Optional[Dict[str, Any]]],
+    segment_detections: List[List[Dict[str, Any]]],
+) -> List[List[Dict[str, Any]]]:
+    # The scores should really be returned from the SAM2 endpoint so we don't have to
+    # try and match them.
+    for detection in transformed_detections:
+        if detection is not None:
+            for i in range(len(detection["scores"])):
+                id_to_score = {}
+                if len(segment_detections) > 0:
+                    # This assumes none of the initial boxes are filtered out by SAM2
+                    # so we have a 1:1 mapping between the initial boxes and the SAM2 boxes
+                    for j, segment_detection in enumerate(segment_detections[0]):
+                        id_to_score[segment_detection["id"]] = detection["scores"][j]
+
+                # after we've created the id_to_score, assign the scores. Some of the
+                # boxes could have been removed in subsequent frames, hence the mapping
+                # is needed
+                for t in range(len(segment_detections)):
+                    for segment_detection in segment_detections[t]:
+                        if segment_detection["id"] in id_to_score:
+                            segment_detection["score"] = id_to_score[
+                                segment_detection["id"]
+                            ]
+                        else:
+                            # if we can't find the score, set it to 1.0 so it doesn't
+                            # get filtered out
+                            segment_detection["score"] = 1.0
+
     return segment_detections
 
 
@@ -141,11 +172,13 @@ def transform_detections(
             bboxes = [
                 denormalize_bbox(detection["bbox"], image_size) for detection in frame
             ]
+            scores = [detection["score"] for detection in frame]
 
             output_list.append(
                 {
                     "labels": labels,
                     "bboxes": bboxes,
+                    "scores": scores,
                 }
             )
         else:
@@ -280,7 +313,7 @@ def post_process(
                     "label": label,
                     "mask": detection["decoded_mask"],
                     "rle": detection["mask"],
-                    "score": 1.0,
+                    "score": detection["score"],
                 }
             )
             del detection["decoded_mask"]

{vision_agent-0.2.235.dist-info → vision_agent-0.2.237.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: vision-agent
-Version: 0.2.235
+Version: 0.2.237
 Summary: Toolset for Vision Agent
 Author: Landing AI
 Author-email: dev@landing.ai
@@ -85,7 +85,7 @@ You can run VisionAgent in a local Jupyter Notebook [Counting cans in an image](
 You can use VisionAgent to generate code to count the number of people in an image:
 ```python
 from vision_agent.agent import VisionAgentCoderV2
-from vision_agent.agent.types import AgentMessage
+from vision_agent.models import AgentMessage
 
 agent = VisionAgentCoderV2(verbose=True)
 code_context = agent.generate_code(

vision_agent-0.2.237.dist-info/RECORD

@@ -0,0 +1,55 @@
+vision_agent/.sim_tools/df.csv,sha256=3tuSr8bjF1pzjEpRJi7LLJssU_2A25SDCWvs4VZKkg4,41479
+vision_agent/.sim_tools/embs.npy,sha256=pi7h3NHlrKncIGNR-oPn_XoTe2PzBb9-aFMi7qK0tEw,245888
+vision_agent/__init__.py,sha256=EAb4-f9iyuEYkBrX4ag1syM8Syx8118_t0R6_C34M9w,57
+vision_agent/agent/README.md,sha256=Q4w7FWw38qaWosQYAZ7NqWx8Q5XzuWrlv7nLhjUd1-8,5527
+vision_agent/agent/__init__.py,sha256=M8CffavdIh8Zh-skznLHIaQkYGCGK7vk4dq1FaVkbs4,617
+vision_agent/agent/agent.py,sha256=RoS7kMfXYILv0zuPpcxqQIlaHGa3K-qw_5EwgsEJTPQ,1530
+vision_agent/agent/vision_agent.py,sha256=4LqvwPTSsiuJEDwBbMx9Dg9ALJwNR6x1c63TZvOMm8A,23486
+vision_agent/agent/vision_agent_coder.py,sha256=Ry6AiyAj3hsSeYPu_5guMcTzf2E4SoebPzpHyJtSPbQ,27360
+vision_agent/agent/vision_agent_coder_prompts.py,sha256=D4RJxTWoxpl-WtYRvHNxaLSdWVHsdYb0jJIQ2ZCGU0A,12277
+vision_agent/agent/vision_agent_coder_prompts_v2.py,sha256=53b_DhQtffX5wxLuCbNQ83AJhB0P_3wEnuKr-v5bx-o,4866
+vision_agent/agent/vision_agent_coder_v2.py,sha256=bWAUyk7-lYKwLIjkL_wUTeYv06zIIroJE1yIDRaGUHw,17059
+vision_agent/agent/vision_agent_planner.py,sha256=rp_atRMDg35WFXNKOTkjUpGPrpSCsiMhcfZtqK-DIV4,18668
+vision_agent/agent/vision_agent_planner_prompts.py,sha256=rYRdJthc-sQN57VgCBKrF09Sd73BSxcBdjNe6C4WNZ8,6837
+vision_agent/agent/vision_agent_planner_prompts_v2.py,sha256=TiiF5BGnFVraFlQnDaeRU67927LvszvpcMUOgVgo0ps,35843
+vision_agent/agent/vision_agent_planner_v2.py,sha256=IqEP5ded5P4ESkLEur81gUvJtTmTdlKAx9uQyLyIwPc,21212
+vision_agent/agent/vision_agent_prompts.py,sha256=KaJwYPUP7_GvQsCPPs6Fdawmi3AQWmWajBUuzj7gTG4,13812
+vision_agent/agent/vision_agent_prompts_v2.py,sha256=Wyxa15NOe75PefAfw3_RRwvgjg8YVqCrU7WvvWoYJpk,2733
+vision_agent/agent/vision_agent_v2.py,sha256=O070_QdgsqNzex5eRtye8QmJgCtHcf_B7zeteWpw3LM,10895
+vision_agent/clients/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+vision_agent/clients/http.py,sha256=k883i6M_4nl7zwwHSI-yP5sAgQZIDPM1nrKD6YFJ3Xs,2009
+vision_agent/clients/landing_public_api.py,sha256=Vz9lldtNbaJRWzT7T8-uQrC-dMnt47LIsDrxHgoVdEw,1492
+vision_agent/configs/__init__.py,sha256=Iu75-w9_nlPmnB_qKA7nYaaaHf7xtTrDmK8N4v2WV34,27
+vision_agent/configs/anthropic_config.py,sha256=T1UuESgiY8913A6wA42P7-cg8FTk9-LkJpyywo7OnIQ,4298
+vision_agent/configs/anthropic_openai_config.py,sha256=YQjFxmlxppn5L55dJjK_v1myBJQ_V5J4q25pmUtwTOU,4310
+vision_agent/configs/config.py,sha256=rUz5zca4Pn5dTUwJXiJzRDYua5PWizApCKI3y0zOvhc,4699
+vision_agent/configs/openai_config.py,sha256=v2_AIY89d7LKWn4uqA2G047U2IdmnqZrGH2Iww9gRIw,4498
+vision_agent/fonts/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+vision_agent/fonts/default_font_ch_en.ttf,sha256=1YM0Z3XqLDjSNbF7ihQFSAIUdjF9m1rtHiNC_6QosTE,1594400
+vision_agent/lmm/__init__.py,sha256=4qX2lmGnKWHeKftXueEi9xj_ieK2nQh_ipHf72nKGFk,84
+vision_agent/lmm/lmm.py,sha256=XYp1frrqQ-6q-0y2IWwM8-EIH5UrFZ21SAhkcM32J9w,19355
+vision_agent/models/__init__.py,sha256=qAdygB-0EsmxMHNzYTPNM6tAF8Fym95gm9bsHJafdgE,287
+vision_agent/models/agent_types.py,sha256=dIdxATH_PP76pD5Wfo0oofWt6iPQh0vpf48QbEQSzhs,2472
+vision_agent/models/lmm_types.py,sha256=v04h-NjbczHOIN8UWa1vvO5-1BDuZ4JQhD2mge1cXmw,305
+vision_agent/models/tools_types.py,sha256=8hYf2OZhI58gvf65KGaeGkt4EQ56nwLFqIQDPHioOBc,2339
+vision_agent/sim/__init__.py,sha256=XYL4BKCB-pamJzCR1y2d5lC8FL64WGu0oEzWzLfguAQ,120
+vision_agent/sim/sim.py,sha256=VSU_1rYd4ifvF45xKWBEYugxdeeEQVpj0QL6rjx49i4,9801
+vision_agent/tools/__init__.py,sha256=T-MPNBVbvWtfo71hobaZsdYzQ52oyymolk_OAb2Pq_g,2463
+vision_agent/tools/meta_tools.py,sha256=-heMwGkx0hX_9zUp1dgBqsJpVnl6Y6tErMsjFy0dwLM,28652
+vision_agent/tools/planner_tools.py,sha256=iXyHjTBIWeQOCfcdQNufoQXfipHu_H38DIoK375FdnA,18492
+vision_agent/tools/prompts.py,sha256=V1z4YJLXZuUl_iZ5rY0M5hHc_2tmMEUKr0WocXKGt4E,1430
+vision_agent/tools/tools.py,sha256=-xg5Msq5ZtHgaISpHnbq5rJ5MIERwfH6wPHg6KpaYjg,111457
+vision_agent/utils/__init__.py,sha256=mANUs_84VL-3gpZbXryvV2mWU623eWnRlJCSUHtMjuw,122
+vision_agent/utils/agent.py,sha256=QGKcbzpAjcVj0958bXYLv07-d2i1GU7-bXVG7bTGRMA,14619
+vision_agent/utils/exceptions.py,sha256=booSPSuoULF7OXRr_YbC4dtKt6gM_HyiFQHBuaW86C4,2052
+vision_agent/utils/execute.py,sha256=vOEP5Ys7S2lc0_7pOJbgk7OaWi85hrCNu9_8Bo3zk6I,29356
+vision_agent/utils/image_utils.py,sha256=bJM2mEvB6E__M9pxi74yQYzAiZ7mu3KE2ptyVrp5vzQ,12533
+vision_agent/utils/tools.py,sha256=USZL0MKsiJgqA8RFiYRTcj_Kn2FVYKLHK4wIk0gP1Ow,7694
+vision_agent/utils/tools_doc.py,sha256=yFue6KSXoa_Z1ngCdBEc4SdPZOWF1rVLeaHu02I8Wis,2523
+vision_agent/utils/type_defs.py,sha256=BE12s3JNQy36QvauXHjwyeffVh5enfcvd4vTzSwvEZI,1384
+vision_agent/utils/video.py,sha256=Dt9_pqGgr63gmpurzisnpF6d9tr65-zxS1CccXdVuxk,6458
+vision_agent/utils/video_tracking.py,sha256=GM9qfeawqhmZVWoKrzw5-NETd4gEo7ImMfWtBnhC3bw,12086
+vision_agent-0.2.237.dist-info/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
+vision_agent-0.2.237.dist-info/METADATA,sha256=MkwC7kWf1f5E1ArMWdjNx_GGNgFwfWQtHbfyDzHN8EM,5755
+vision_agent-0.2.237.dist-info/WHEEL,sha256=7Z8_27uaHI_UZAc4Uox4PpBhQ9Y5_modZXWMxtUi4NU,88
+vision_agent-0.2.237.dist-info/RECORD,,

vision_agent-0.2.235.dist-info/RECORD

@@ -1,52 +0,0 @@
-vision_agent/.sim_tools/df.csv,sha256=oVUuyoVTCnayorbGUAvWed8l1YA_-rF9rSF78fMtvuU,42468
-vision_agent/.sim_tools/embs.npy,sha256=YJe8EcKVNmeX_75CS2T1sbY-sUS_1HQAMT-34zc18a0,254080
-vision_agent/__init__.py,sha256=EAb4-f9iyuEYkBrX4ag1syM8Syx8118_t0R6_C34M9w,57
-vision_agent/agent/README.md,sha256=Q4w7FWw38qaWosQYAZ7NqWx8Q5XzuWrlv7nLhjUd1-8,5527
-vision_agent/agent/__init__.py,sha256=M8CffavdIh8Zh-skznLHIaQkYGCGK7vk4dq1FaVkbs4,617
-vision_agent/agent/agent.py,sha256=_1tHWAs7Jm5tqDzEcPfCRvJV3uRRveyh4n9_9pd6I1w,1565
-vision_agent/agent/agent_utils.py,sha256=4RgG8SUEGuMFHkIt0jCFkRQF6G1PZp3Ub4LuVYKF7Ic,14092
-vision_agent/agent/types.py,sha256=dIdxATH_PP76pD5Wfo0oofWt6iPQh0vpf48QbEQSzhs,2472
-vision_agent/agent/vision_agent.py,sha256=fH9NOLk7twL1fPr9vLSqkaYhah-gfDWfTOVF2FfMyzI,23461
-vision_agent/agent/vision_agent_coder.py,sha256=flUxOibyGZK19BCSK5mhaD3HjCxHw6c6FtKom6N2q1E,27359
-vision_agent/agent/vision_agent_coder_prompts.py,sha256=_kkPLezUVnBXieNPlxMQab_6J6P7F-aa6ItF5NhZZsM,12281
-vision_agent/agent/vision_agent_coder_prompts_v2.py,sha256=NUMWq-Lxq5JmmyWs3C5O_1Hm-zCbf9I_yPK5UtWGspE,4871
-vision_agent/agent/vision_agent_coder_v2.py,sha256=yQYcO0s4BI9pWaAQQAVtkwWa3UF5w0iLKvwpeJ6iegM,17077
-vision_agent/agent/vision_agent_planner.py,sha256=fFzjNkZBKkh8Y_oS06ATI4qz31xmIJvixb_tV1kX8KA,18590
-vision_agent/agent/vision_agent_planner_prompts.py,sha256=rYRdJthc-sQN57VgCBKrF09Sd73BSxcBdjNe6C4WNZ8,6837
-vision_agent/agent/vision_agent_planner_prompts_v2.py,sha256=JLiFZvwQFLYukgj4l-SzxSIjmJHTEkL2HSZbkjU529w,35591
-vision_agent/agent/vision_agent_planner_v2.py,sha256=wISmdTN-W1pjgZg3_aneGowI3lRQRHTSbpyeTJ79O5A,20645
-vision_agent/agent/vision_agent_prompts.py,sha256=KaJwYPUP7_GvQsCPPs6Fdawmi3AQWmWajBUuzj7gTG4,13812
-vision_agent/agent/vision_agent_prompts_v2.py,sha256=Wyxa15NOe75PefAfw3_RRwvgjg8YVqCrU7WvvWoYJpk,2733
-vision_agent/agent/vision_agent_v2.py,sha256=Q96YsLovCUOd6m2Cg7EGCiHshDq65vxHsfVj7IToyls,10936
-vision_agent/clients/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-vision_agent/clients/http.py,sha256=k883i6M_4nl7zwwHSI-yP5sAgQZIDPM1nrKD6YFJ3Xs,2009
-vision_agent/clients/landing_public_api.py,sha256=lU2ev6E8NICmR8DMUljuGcVFy5VNJQ4WQkWC8WnnJEc,1503
-vision_agent/configs/__init__.py,sha256=Iu75-w9_nlPmnB_qKA7nYaaaHf7xtTrDmK8N4v2WV34,27
-vision_agent/configs/anthropic_config.py,sha256=T1UuESgiY8913A6wA42P7-cg8FTk9-LkJpyywo7OnIQ,4298
-vision_agent/configs/anthropic_openai_config.py,sha256=YQjFxmlxppn5L55dJjK_v1myBJQ_V5J4q25pmUtwTOU,4310
-vision_agent/configs/config.py,sha256=YQjFxmlxppn5L55dJjK_v1myBJQ_V5J4q25pmUtwTOU,4310
-vision_agent/configs/openai_config.py,sha256=v2_AIY89d7LKWn4uqA2G047U2IdmnqZrGH2Iww9gRIw,4498
-vision_agent/fonts/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-vision_agent/fonts/default_font_ch_en.ttf,sha256=1YM0Z3XqLDjSNbF7ihQFSAIUdjF9m1rtHiNC_6QosTE,1594400
-vision_agent/lmm/__init__.py,sha256=xk2Rn8Zgpy2xwYaOGHzy4tXxnxo2aj6SkpNjeJ8yxcY,111
-vision_agent/lmm/lmm.py,sha256=arwfYPWme_RxCxSpEQ0ZkpHO22GFPCwVeoSvXqLPOAk,19288
-vision_agent/lmm/types.py,sha256=ZEXR_ptBL0ZwDMTDYkgxUCmSZFmBYPQd2jreNzr_8UY,221
-vision_agent/tools/__init__.py,sha256=zopUrANPx7p0NGy6BxmEaYhDrj8DX8w7BLfgmCbz-mU,2897
-vision_agent/tools/meta_tools.py,sha256=TPeS7QWnc_PmmU_ndiDT03dXbQ5yDSP33E7U8cSj7Ls,28660
-vision_agent/tools/planner_tools.py,sha256=8pJZCGGOGIqGiV2or52BjyRP6eDlporuQ2hXCIHfLTQ,15382
-vision_agent/tools/prompts.py,sha256=V1z4YJLXZuUl_iZ5rY0M5hHc_2tmMEUKr0WocXKGt4E,1430
-vision_agent/tools/tool_utils.py,sha256=l4oWkgPd_s8QzXqqbrLwPgcfnhsJaPpdMKikOcwRaoQ,10396
-vision_agent/tools/tools.py,sha256=MBeFVYyCx-QQLCFb0Cn4m6SgmT6-6HxUOmCKAARrv6s,108547
-vision_agent/tools/tools_types.py,sha256=8hYf2OZhI58gvf65KGaeGkt4EQ56nwLFqIQDPHioOBc,2339
-vision_agent/utils/__init__.py,sha256=QKk4zVjMwGxQI0MQ-aZZA50N-qItxRY4EB9CwQkZ2HY,185
-vision_agent/utils/exceptions.py,sha256=booSPSuoULF7OXRr_YbC4dtKt6gM_HyiFQHBuaW86C4,2052
-vision_agent/utils/execute.py,sha256=vOEP5Ys7S2lc0_7pOJbgk7OaWi85hrCNu9_8Bo3zk6I,29356
-vision_agent/utils/image_utils.py,sha256=z_ONgcza125B10NkoGwPOzXnL470bpTWZbkB16NeeH0,12188
-vision_agent/utils/sim.py,sha256=DYya76dYVtifFyXilMLxBzGgyfyeqhEwU4RJ4894lCI,9796
-vision_agent/utils/type_defs.py,sha256=BE12s3JNQy36QvauXHjwyeffVh5enfcvd4vTzSwvEZI,1384
-vision_agent/utils/video.py,sha256=e1VwKhXzzlC5LcFMyrcQYrPnpnX4wxDpnQ-76sB4jgM,6001
-vision_agent/utils/video_tracking.py,sha256=PXZYB0ZJM97WU52XbucyoXX2GW9-gNpSHrPF30_Lq1Q,10263
-vision_agent-0.2.235.dist-info/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
-vision_agent-0.2.235.dist-info/METADATA,sha256=bAPsRs8veydvltpYRLYxAOB_bhASv9zKkE9TjkPVm2Q,5760
-vision_agent-0.2.235.dist-info/WHEEL,sha256=7Z8_27uaHI_UZAc4Uox4PpBhQ9Y5_modZXWMxtUi4NU,88
-vision_agent-0.2.235.dist-info/RECORD,,