Remove matplotlib from conda requirements.

conda-build/habitat-sim/meta.yaml

@@ -44,7 +44,6 @@ requirements:
     - tqdm
     - imageio
     - imageio-ffmpeg
-    - matplotlib
     - libglu                             # [linux]
     - libxcb                             # [linux]
     - xorg-libx11                        # [linux]
@@ -69,7 +68,7 @@ requirements:
     - scipy>=1.3.0
     - tqdm
     - imageio
-    - matplotlib
+    - imageio-ffmpeg
     - libxcb                             # [linux]
     - xorg-libx11                        # [linux]
     - xorg-libxau                        # [linux]

examples/tutorials/lighting_tutorial.py

@@ -8,7 +8,7 @@
 
 import magnum as mn
 import numpy as np
-from matplotlib import pyplot as plt
+from PIL import Image
 
 import habitat_sim
 from habitat_sim.gfx import LightInfo, LightPositionModel
@@ -22,20 +22,17 @@
 
 
 def show_img(data, save):
-    plt.figure(figsize=(12, 12))
-    plt.imshow(data, interpolation="nearest")
-    plt.axis("off")
-    plt.show(block=False)
+    # Convert the data to a PIL Image
+    img = Image.fromarray((data).astype(np.uint8))
+    # Convert to RGB if the image has an alpha channel
+    if img.mode == "RGBA":
+        img = img.convert("RGB")
+    # Display the image
+    img.show()
     if save:
         global save_index
-        plt.savefig(
-            output_path + str(save_index) + ".jpg",
-            bbox_inches="tight",
-            pad_inches=0,
-            quality=50,
-        )
+        img.save(output_path + str(save_index) + ".jpg", quality=50)
         save_index += 1
-    plt.pause(1)
 
 
 def get_obs(sim, show, save):

examples/tutorials/nb_python/ECCV_2020_Advanced_Features.py

@@ -599,7 +599,7 @@ def restore_camera_track_config(sim, init_state):
     visual_sensor._spec.position = init_state["position"]
     visual_sensor._spec.orientation = init_state["orientation"]
     visual_sensor._sensor_object.set_transformation_from_spec()
-    # restore the agent's state to what was savedd in init_camera_track_config
+    # restore the agent's state to what was saved in init_camera_track_config
     sim.get_agent(agent_ID).set_state(init_state["agent_state"])
 
 
@@ -1016,7 +1016,7 @@ def get_2d_point(sim, sensor_name, point_3d):
 make_simulator_from_settings(sim_settings)
 
 # fmt: off
-# @markdown In this example, we load a box asset with each face as a separate component with its own SceneNode. We demonstrate the result of modiyfing the associated semantic ids via object templates, the Simulator API, and the SceneNode property.
+# @markdown In this example, we load a box asset with each face as a separate component with its own SceneNode. We demonstrate the result of modifying the associated semantic ids via object templates, the Simulator API, and the SceneNode property.
 # fmt: on
 
 rigid_obj_mgr.remove_all_objects()
@@ -1262,7 +1262,7 @@ def get_2d_point(sim, sensor_name, point_3d):
 force_flat_shading = True  # @param {type:"boolean"}
 new_template.force_flat_shading = force_flat_shading
 
-# @markdown The x,y,z components of the intertia matrix diagonal
+# @markdown The x,y,z components of the inertia matrix diagonal
 
 inertia_X = 1.0  # @param {type:"slider", min:0.1, max:10, step:0.1}
 inertia_Y = 1  # @param {type:"slider", min:0.1, max:10, step:0.1}
@@ -1398,7 +1398,7 @@ def get_2d_point(sim, sensor_name, point_3d):
 # @markdown Note 4 : There often are different modifiable properties available for solid and wireframe versions of the same primitive.
 
 # @markdown ###Primitive Asset Template Properties
-# @markdown The flollowing examples will illustrate all the modifiable properties for each of the different types of primitives, and allow for their synthesis.
+# @markdown The following examples will illustrate all the modifiable properties for each of the different types of primitives, and allow for their synthesis.
 
 
 # Primitive Asset Attributes Manager, provides access to AssetAttributesTemplates

examples/tutorials/notebooks/ECCV_2020_Advanced_Features.ipynb

@@ -605,7 +605,7 @@
     "    visual_sensor._spec.position = init_state[\"position\"]\n",
     "    visual_sensor._spec.orientation = init_state[\"orientation\"]\n",
     "    visual_sensor._sensor_object.set_transformation_from_spec()\n",
-    "    # restore the agent's state to what was savedd in init_camera_track_config\n",
+    "    # restore the agent's state to what was saved in init_camera_track_config\n",
     "    sim.get_agent(agent_ID).set_state(init_state[\"agent_state\"])\n",
     "\n",
     "\n",
@@ -1080,7 +1080,7 @@
     "make_simulator_from_settings(sim_settings)\n",
     "\n",
     "# fmt: off\n",
-    "# @markdown In this example, we load a box asset with each face as a separate component with its own SceneNode. We demonstrate the result of modiyfing the associated semantic ids via object templates, the Simulator API, and the SceneNode property.\n",
+    "# @markdown In this example, we load a box asset with each face as a separate component with its own SceneNode. We demonstrate the result of modifying the associated semantic ids via object templates, the Simulator API, and the SceneNode property.\n",
     "# fmt: on\n",
     "\n",
     "rigid_obj_mgr.remove_all_objects()\n",
@@ -1363,7 +1363,7 @@
     "force_flat_shading = True  # @param {type:\"boolean\"}\n",
     "new_template.force_flat_shading = force_flat_shading\n",
     "\n",
-    "# @markdown The x,y,z components of the intertia matrix diagonal\n",
+    "# @markdown The x,y,z components of the inertia matrix diagonal\n",
     "\n",
     "inertia_X = 1.0  # @param {type:\"slider\", min:0.1, max:10, step:0.1}\n",
     "inertia_Y = 1  # @param {type:\"slider\", min:0.1, max:10, step:0.1}\n",
@@ -1522,7 +1522,7 @@
     "# @markdown Note 4 : There often are different modifiable properties available for solid and wireframe versions of the same primitive.\n",
     "\n",
     "# @markdown ###Primitive Asset Template Properties\n",
-    "# @markdown The flollowing examples will illustrate all the modifiable properties for each of the different types of primitives, and allow for their synthesis.\n",
+    "# @markdown The following examples will illustrate all the modifiable properties for each of the different types of primitives, and allow for their synthesis.\n",
     "\n",
     "\n",
     "# Primitive Asset Attributes Manager, provides access to AssetAttributesTemplates\n",

examples/tutorials/semantic_id_tutorial.py

@@ -8,7 +8,7 @@
 
 import magnum as mn
 import numpy as np
-from matplotlib import pyplot as plt
+from PIL import Image
 
 import habitat_sim
 from habitat_sim.utils.common import quat_from_angle_axis
@@ -20,27 +20,44 @@
 save_index = 0
 
 
+def apply_colormap(image):
+    # Normalize the uint32 image to the range [0, 255]
+    normalized = (image - image.min()) / (image.max() - image.min()) * 255
+    normalized = normalized.astype(np.uint8)
+
+    # Create a color map
+    colormap = np.zeros((256, 3), dtype=np.uint8)
+    for i in range(256):
+        colormap[i] = [i, 255 - i, (i // 2) % 256]
+
+    # Apply the colormap
+    colorized = colormap[normalized]
+    return Image.fromarray(colorized)
+
+
 def show_img(data, save):
-    # display rgb and semantic images side-by-side
-    fig = plt.figure(figsize=(12, 12))
-    ax1 = fig.add_subplot(1, 2, 1)
-    ax1.axis("off")
-    ax1.imshow(data[0], interpolation="nearest")
-    ax2 = fig.add_subplot(1, 2, 2)
-    ax2.axis("off")
-    ax2.imshow(data[1], interpolation="nearest")
-    plt.axis("off")
-    plt.show(block=False)
+    # Convert numpy arrays to PIL Images
+    img1 = Image.fromarray(data[0])
+
+    # Color-code the semantic map (uint32) to visualize the semantic IDs
+    img2 = apply_colormap(data[1])
+
+    # Create a new image with twice the width to place images side-by-side
+    total_width = img1.width + img2.width
+    max_height = max(img1.height, img2.height)
+    new_img = Image.new("RGB", (total_width, max_height))
+
+    # Paste the images side-by-side
+    new_img.paste(img1, (0, 0))
+    new_img.paste(img2, (img1.width, 0))
+
+    # Display the combined image
+    new_img.show()
+
     if save:
         global save_index
-        plt.savefig(
-            output_path + str(save_index) + ".jpg",
-            bbox_inches="tight",
-            pad_inches=0,
-            quality=50,
-        )
+        new_img.save(output_path + str(save_index) + ".jpg", quality=50)
         save_index += 1
-    plt.pause(1)
 
 
 def get_obs(sim, show, save):

requirements.txt

@@ -2,7 +2,6 @@ attrs>=19.1.0
 gitpython
 imageio
 imageio-ffmpeg
-matplotlib
 numba
 numpy==1.26.4
 numpy-quaternion