utils.py<\/code> file and get started.<\/p>\nimport matplotlib.pyplot as plt\nimport numpy as np\n\n\ndef show_points(coords, labels, ax, marker_size=375):\n pos_points = coords[labels == 1]\n neg_points = coords[labels == 0]\n ax.scatter(\n pos_points[:, 0],\n pos_points[:, 1],\n color=\"green\",\n marker=\"*\",\n s=marker_size,\n edgecolor=\"white\",\n linewidth=1.25,\n )\n ax.scatter(\n neg_points[:, 0],\n neg_points[:, 1],\n color=\"red\",\n marker=\"*\",\n s=marker_size,\n edgecolor=\"white\",\n linewidth=1.25,\n )\n\n\ndef show_box(box, ax, processed_dim=False):\n x0, y0 = box[0], box[1]\n if processed_dim:\n w, h = box[2], box[3]\n else:\n w, h = box[2] - box[0], box[3] - box[1]\n ax.add_patch(\n plt.Rectangle((x0, y0), w, h, edgecolor=\"green\", facecolor=(0, 0, 0, 0), lw=2)\n )\n\n\ndef show_mask(mask, ax):\n color = np.array([0, 0, 1, 0.6])\n h, w = mask.shape[-2:]\n mask_image = mask.reshape(h, w, 1) * color.reshape(1, 1, -1)\n ax.imshow(mask_image)\n\n\ndef show_all_masks(masks):\n sorted_masks = sorted(masks, key=(lambda x: x[\"area\"]), reverse=True)\n ax = plt.gca()\n ax.set_autoscale_on(False)\n\n img = np.ones(\n (\n sorted_masks[0][\"segmentation\"].shape[0],\n sorted_masks[0][\"segmentation\"].shape[1],\n 4,\n )\n )\n img[:, :, 3] = 0\n for mask in sorted_masks:\n m = mask[\"segmentation\"]\n color_mask = np.concatenate([np.random.random(3), [0.35]])\n img[m] = color_mask\n ax.imshow(img)<\/pre>\nWe start by importing the necessary packages, such as matplotlib<\/code> (Line 1<\/strong>) and numpy<\/code> (Line 2<\/strong>).<\/p>\nNext, we define our helper functions which will allow us to plot the different prompts and visualize the segmentation masks.<\/p>\n
We start by implementing the show_points<\/code> function (Lines 5-25<\/strong>), which allows us to visualize the points in the prompt by plotting them on the figure input to the function.<\/p>\nThe function takes as input the point coordinates, binary labels corresponding to the points (indicating whether we want the region with or without the point in the predicted segmentation mask), the matplotlib figure parameter (i.e., ax<\/code>), and the marker_size<\/code> for the plotted point (Line 5<\/strong>).<\/p>\nNext, we get the points with labels equal to 1<\/code>, store them as pos_points<\/code>, and also get the points with labels equal to 0<\/code> and store them as neg_points<\/code> (Lines 6 and 7<\/strong>).<\/p>\nThen we use the matplotlib scatter()<\/code> function to plot the pos_points<\/code> with green color and the neg_points<\/code> with red color, as shown on Lines 8-25<\/strong>. Note that this function takes as input the x<\/code> and y<\/code> coordinates of the points, the marker type, the color for the marker to plot, the size of the marker, the color of the edge, and the width of line as shown.<\/p>\nNext, we define the show_box<\/code> function, which allows us to visualize the bounding box in the prompt by plotting it on the figure input to the function.<\/p>\nThe function takes as input the box (i.e., a list with coordinates of the box), the matplotlib figure parameter (i.e., ax<\/code>) on which we want to plot the box and the processed_dim<\/code> argument.<\/p>\nNote that the input box provided as argument (which is a set of 4 values) can be in either of 2 formats, that is, <\/p>\n
\n- Case 1:<\/strong>
x, y<\/code> coordinates of bottom left corner and x, y<\/code> coordinates of the top right corner <\/li>\n- Case 2:<\/strong>
x, y<\/code> coordinates of bottom left corner and the width and height of the box.<\/li>\n<\/ul>\nThe processed_dim<\/code> argument states which format the bounding box is provided in. It is False<\/code> in Case 1<\/strong> and True<\/code> in Case 2<\/strong>.<\/p>\nOn Line 29<\/strong>, we get the x0<\/code>, y0<\/code> coordinates as shown. In case the processed_dim<\/code> flag is True<\/code>, we get the width and height of the bounding box (i.e., box[2]<\/code> and box[3]<\/code>), as shown on Line<\/strong> 31<\/strong>. If not, we compute the width and height of the bounding box as follows, w=box[2]-box[0]<\/code> and h=box[3]-box[1]<\/code>), as shown on line Line 33<\/strong>. <\/p>\nFinally, we use the add_patch<\/code> function from matplotlib and plot a green-colored rectangle using the plt.Rectangle<\/code> function, as shown on Lines 34-36<\/strong>.<\/p>\nNow that we have implemented functions to visualize the prompts, we define the functions that will allow us to visualize the segmentation masks.<\/p>\n
We start with the show_mask<\/code> function, which takes as input a single predicted mask, the matplotlib figure parameter (i.e., ax<\/code>) on which we want to plot and visualize the mask.<\/p>\nOn Line 40<\/strong>, we define the color, which is an array with the R, G, B value and the value of transparency with which the mask is plotted on the image (i.e., [0, 0, 1, 0.6]<\/code>). Next, we get the height and width of the input mask (Line 41<\/strong>).<\/p>\nNext, we reshape the mask to have the shape in the format (h,w,1)<\/code>, reshape the color to have shape (1,1,4)<\/code>, and multiply both, as shown on Line 42<\/strong>. This broadcasts the mask to the shape (h,w,4)<\/code> and color to the shape (h,w,4)<\/code>, which gets multiplied elementwise.<\/p>\nWe finally use the imshow()<\/code> function to visualize the mask_image<\/code>, as shown on
![\"\"](\"https:\/\/b2633864.smushcdn.com\/2633864\/wp-content\/uploads\/2023\/09\/sam-part1-featured-1024x575.png?lossy=2&strip=1&webp=1\")
<\/a><\/figure>\n<\/div>\n![\"\"](\"https:\/\/b2633864.smushcdn.com\/2633864\/wp-content\/uploads\/2020\/01\/source-code-icon.png?lossy=2&strip=1&webp=1\"){kind=icon}
<\/p>\n