Update README.md

README.md

@@ -1,27 +1,48 @@
----
-license: mit
----
-git clone https://github.com/cliport/cliport.git
-from susie.model import create_sample_fn
-from susie.jax_utils import initialize_compilation_cache
-import requests
+from itertools import product, chain
+import math
+import matplotlib.pyplot as plt
 import numpy as np
-from PIL import Image
-https://github.com/cliport/cliport/releases/download/v1.0.0/cliport_quickstart.zip
-initialize_compilation_cache()
+import networkx as nx
+import cv2
+im = cv2.imread("/kaggle/working/black.png")
+im = cv2.cvtColor(im, cv2.COLOR_BGR2GRAY) 
 
-IMAGE_URL = "https://rail.eecs.berkeley.edu/datasets/bridge_release/raw/bridge_data_v2/datacol2_toykitchen7/drawer_pnp/01/2023-04-19_09-18-15/raw/traj_group0/traj0/images0/im_12.jpg"
+h, w = im.shape # Get height and width of image
 
-sample_fn = create_sample_fn("kvablack/susie")
-image = np.array(Image.open(requests.get(IMAGE_URL, stream=True).raw).resize((256, 256)))
-image_out = sample_fn(image, "open the drawer")
+# Add only those nodes which are black to graph
+nodes = [(i, j) for (i, j) in product(range(h), range(w)) if im[i, j] == 0]
+g = nx.Graph(nodes)
 
-# to display the images if you're in a Jupyter notebook
-display(Image.fromarray(image))
-display(Image.fromarray(image_out))
+# For each node there can be 8 neighbours, if you consider diagonal as well.
+def get_neighbors(node):
+    box_coords = product([-1, 0, 1], [-1, 0, 1])
+    nns = []
+    for coord in box_coords:
+        if coord[0] != coord[1]:
+            nn = (node[0] - coord[0], node[1] - coord[1])
+            nns.append(nn)
+    return nns
 
-https://github.com/kvablack/susie/tree/8177f63332a3905202c122310a92c51b8ff14280
-2nd one
-Imagine you are a human, how would you break these visual tasks down to step by step?I will give you a task, some objects and you must break them down step by step, each step should be roughly 5 words but correct. 
-Task: clean the window with the sponge and then put it down in the table
-Objects: Window, table, sponge near table
+# A point will be a neighbour if it is black as well and is in image bounds
+neighbors = list(chain.from_iterable([[(node, ng_node, 1) for ng_node in  get_neighbors(node) if (im[node] == 0) and (0 < ng_node[0] < h) and (0 < ng_node[1] , w)] for node in nodes]))
+
+g.add_weighted_edges_from(neighbors)
+
+# In image loaded above (0, 0) is top left point. To keep things little more generic. I select point closest to (0, 0) as start and furthest as end.
+
+min_pt = min(nodes, key=lambda x: math.hypot(x[0], x[1]))
+max_pt = max(nodes, key=lambda x: math.hypot(x[0], x[1]))
+
+# Now we can just use networkx to find path between two points
+path = nx.shortest_path(g, source=min_pt, target=max_pt)
+
+# Get new image with only shortest path
+im2 = 255*np.ones_like(im)
+# Draw the path on the original image before saving
+# Draw the path on the original image with white color
+for i in range(len(path)-1):
+    cv2.line(im, path[i], path[i+1], color=(255), thickness=1)
+
+cv2.imwrite('image_with_path.png', im)
+plt.figure(figsize=(10, 10))
+plt.imshow(im, cmap='gray')