Add computer vision sample

Author: saihv

grid-samples/vqa_seg.py

@@ -0,0 +1,39 @@
+import os
+import rerun as rr
+import airgen
+from airgen.utils.sensor import ImageType, imagetype2request, responses2images
+
+# Enable logging visual data inside GRID's models
+os.environ["TURN_ON_RERUN"] = "1"
+
+from grid import GRIDConfig
+# Replace this with the main directory of GRID that contains `external`
+GRIDConfig.set_main_dir("/workspaces/GRID")
+
+from grid.model.perception.vqa.gllava import GLLaVA
+from grid.model.perception.segmentation.gsam import GroundedSAM
+
+llava = GLLaVA()
+gsam = GroundedSAM()
+
+# Initialize the AirGen client
+c = airgen.MultirotorClient()
+c.confirmConnection()
+c.enableApiControl(True)
+c.armDisarm(True)
+c.takeoffAsync().join() 
+
+responses = c.simGetImages([imagetype2request("front_center", ImageType.Scene)])
+image = responses2images(responses)[0][0]
+obj = llava.run(image, "Describe the scene, and pick a single object of interest")
+print(obj)
+
+# Example output: The scene is a large, open field with a road running through it. 
+#                   In the distance, there are several wind turbines, which are the main focus 
+#                   of the image. These wind turbines are situated in the middle of the field, 
+#                   and they are spinning, indicating that they are generating electricity from 
+#                   the wind. The presence of these turbines suggests that the area is likely 
+#                   a wind farm, which harnesses the power of the wind to produce clean and 
+#                   renewable energy
+
+seg = gsam.segment_object(image, "wind turbines")