vision-agent 0.2.44__py3-none-any.whl → 0.2.45__py3-none-any.whl

vision_agent/agent/vision_agent.py

@@ -2,9 +2,11 @@ import copy
 import json
 import logging
 import sys
+import tempfile
 from pathlib import Path
-from typing import Any, Callable, Dict, List, Optional, Union, cast
+from typing import Any, Callable, Dict, List, Optional, Sequence, Union, cast
 
+from PIL import Image
 from rich.console import Console
 from rich.style import Style
 from rich.syntax import Syntax
@@ -78,12 +80,35 @@ def extract_json(json_str: str) -> Dict[str, Any]:
     return json_dict  # type: ignore
 
 
+def extract_image(
+    media: Optional[Sequence[Union[str, Path]]]
+) -> Optional[Sequence[Union[str, Path]]]:
+    if media is None:
+        return None
+
+    new_media = []
+    for m in media:
+        m = Path(m)
+        extension = m.suffix
+        if extension in [".jpg", ".jpeg", ".png", ".bmp"]:
+            new_media.append(m)
+        elif extension in [".mp4", ".mov"]:
+            frames = T.extract_frames(m)
+            with tempfile.NamedTemporaryFile(suffix=".png", delete=False) as tmp:
+                if len(frames) > 0:
+                    Image.fromarray(frames[0][0]).save(tmp.name)
+                    new_media.append(Path(tmp.name))
+    if len(new_media) == 0:
+        return None
+    return new_media
+
+
 def write_plan(
     chat: List[Dict[str, str]],
     tool_desc: str,
     working_memory: str,
     model: Union[LLM, LMM],
-    media: Optional[List[Union[str, Path]]] = None,
+    media: Optional[Sequence[Union[str, Path]]] = None,
 ) -> List[Dict[str, str]]:
     chat = copy.deepcopy(chat)
     if chat[-1]["role"] != "user":
@@ -94,6 +119,7 @@ def write_plan(
     prompt = PLAN.format(context=context, tool_desc=tool_desc, feedback=working_memory)
     chat[-1]["content"] = prompt
     if isinstance(model, OpenAILMM):
+        media = extract_image(media)
         return extract_json(model.chat(chat, images=media))["plan"]  # type: ignore
     else:
         return extract_json(model.chat(chat))["plan"]  # type: ignore
@@ -103,7 +129,7 @@ def reflect(
     chat: List[Dict[str, str]],
     plan: str,
     code: str,
-    model: LLM,
+    model: Union[LLM, LMM],
 ) -> Dict[str, Union[str, bool]]:
     chat = copy.deepcopy(chat)
     if chat[-1]["role"] != "user":
@@ -309,7 +335,7 @@ class VisionAgent(Agent):
 
     def __init__(
         self,
-        planner: Optional[LLM] = None,
+        planner: Optional[Union[LLM, LMM]] = None,
         coder: Optional[LLM] = None,
         tester: Optional[LLM] = None,
         debugger: Optional[LLM] = None,

vision_agent/agent/vision_agent_prompts.py

@@ -29,14 +29,17 @@ PLAN = """
 {feedback}
 
 **Instructions**:
-Based on the context and tools you have available, write a plan of subtasks to achieve the user request utilizing given tools when necessary. Output a list of jsons in the following format:
+1. Based on the context and tools you have available, write a plan of subtasks to achieve the user request.
+2. Go over the users request step by step and ensure each step is represented as a clear subtask in your plan.
+
+Output a list of jsons in the following format
 
 ```json
 {{
     "plan":
         [
             {{
-                "instructions": str # what you should do in this task, one short phrase or sentence
+                "instructions": str # what you should do in this task associated with a tool
             }}
         ]
 }}

vision_agent/tools/tools.py

@@ -199,14 +199,15 @@ def extract_frames(
 
 def ocr(image: np.ndarray) -> List[Dict[str, Any]]:
     """'ocr' extracts text from an image. It returns a list of detected text, bounding
-    boxes, and confidence scores. The results are sorted from top-left to bottom right
+    boxes with normalized coordinates, and confidence scores. The results are sorted
+    from top-left to bottom right.
 
     Parameters:
         image (np.ndarray): The image to extract text from.
 
     Returns:
-        List[Dict[str, Any]]: A list of dictionaries containing the detected text, bbox,
-            and confidence score.
+        List[Dict[str, Any]]: A list of dictionaries containing the detected text, bbox
+            with nornmalized coordinates, and confidence score.
 
     Example
     -------

{vision_agent-0.2.44.dist-info → vision_agent-0.2.45.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: vision-agent
-Version: 0.2.44
+Version: 0.2.45
 Summary: Toolset for Vision Agent
 Author: Landing AI
 Author-email: dev@landing.ai

{vision_agent-0.2.44.dist-info → vision_agent-0.2.45.dist-info}/RECORD

@@ -11,8 +11,8 @@ vision_agent/agent/easytool_v2.py,sha256=CjY-sSj3abxnSq3ZHZMt-7YvRWDXEZsC6RN8FFI
 vision_agent/agent/easytool_v2_prompts.py,sha256=MZSIwovYgB-f-kdJ6btaNDVXptJn47bfOL3-Zn6NiC0,8573
 vision_agent/agent/reflexion.py,sha256=AlM5AvBJvCslXlYQdZiadq4oVHsNBm3IF_03DglTxRo,10506
 vision_agent/agent/reflexion_prompts.py,sha256=G7UAeNz_g2qCb2yN6OaIC7bQVUkda4m3z42EG8wAyfE,9342
-vision_agent/agent/vision_agent.py,sha256=JtPDIiLINXm3jBR0LbqblfB9yCv-8M-B7XRx1EPDhFU,18749
-vision_agent/agent/vision_agent_prompts.py,sha256=FnIYF2Fe3joRvFnOJD9ZyWXMihMyL606nXxWJ0adTZ8,8314
+vision_agent/agent/vision_agent.py,sha256=c3jJd1uiCtmVC2xazUvW9rwc7usi-EOYW7NZnMFOdt8,19586
+vision_agent/agent/vision_agent_prompts.py,sha256=bIcqutsyM2bEhWE2XGw01PuZ9f-jePSwapbvkOOrFZ4,8384
 vision_agent/fonts/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 vision_agent/fonts/default_font_ch_en.ttf,sha256=1YM0Z3XqLDjSNbF7ihQFSAIUdjF9m1rtHiNC_6QosTE,1594400
 vision_agent/llm/__init__.py,sha256=BoUm_zSAKnLlE8s-gKTSQugXDqVZKPqYlWwlTLdhcz4,48
@@ -23,14 +23,14 @@ vision_agent/tools/__init__.py,sha256=K_7knxmyTIcSEGL8c9wF8RpVh3GrMYfybFaq-2SUM1
 vision_agent/tools/easytool_tools.py,sha256=pZc5dQlYINlV4nYbbzsDi3-wauA-fCeD2iGmJUMoUfE,47373
 vision_agent/tools/prompts.py,sha256=V1z4YJLXZuUl_iZ5rY0M5hHc_2tmMEUKr0WocXKGt4E,1430
 vision_agent/tools/tool_utils.py,sha256=wzRacbUpqk9hhfX_Y08rL8qP0XCN2w-8IZoYLi3Upn4,869
-vision_agent/tools/tools.py,sha256=PhmJ0kQeZ-tSQ675HI8QnR49zlH6nJ_opt6QS4dNSVA,25889
+vision_agent/tools/tools.py,sha256=66pFXUIVvnOa1fk0PY5u_75kblIbAVqkRP2U9qLixrY,25951
 vision_agent/utils/__init__.py,sha256=Ce4yPhoWanRsnTy3X7YzZNBYYRJsrJeT7N59WUf8GZM,209
 vision_agent/utils/execute.py,sha256=GlpUGe3pg5KdSvRHLFfVcn9ptXBIp-QRoHT3Wa6aIMs,20318
 vision_agent/utils/image_utils.py,sha256=_cdiS5YrLzqkq_ZgFUO897m5M4_SCIThwUy4lOklfB8,7700
 vision_agent/utils/sim.py,sha256=oUZ-6eu8Io-UNt9GXJ0XRKtP-Wc0sPWVzYGVpB2yDFk,3001
 vision_agent/utils/type_defs.py,sha256=BlI8ywWHAplC7kYWLvt4AOdnKpEW3qWEFm-GEOSkrFQ,1792
 vision_agent/utils/video.py,sha256=EuJJ7Owi3pIV-q3WcZ-LaaTrGAmmZ8YAA22rmEkY7GI,8885
-vision_agent-0.2.44.dist-info/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
-vision_agent-0.2.44.dist-info/METADATA,sha256=EbnJiKZzbAgeCN30GRMYfMPN5w_wo9XBkuhWEP_0cN8,6817
-vision_agent-0.2.44.dist-info/WHEEL,sha256=7Z8_27uaHI_UZAc4Uox4PpBhQ9Y5_modZXWMxtUi4NU,88
-vision_agent-0.2.44.dist-info/RECORD,,
+vision_agent-0.2.45.dist-info/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
+vision_agent-0.2.45.dist-info/METADATA,sha256=G9Cy7cUPEWi42cuVP8V7u_ZjNXEnYp26_kb_u9rXSQk,6817
+vision_agent-0.2.45.dist-info/WHEEL,sha256=7Z8_27uaHI_UZAc4Uox4PpBhQ9Y5_modZXWMxtUi4NU,88
+vision_agent-0.2.45.dist-info/RECORD,,