PSG / utils.py
liangyuch's picture
Update utils.py
809c0a6
from typing import Tuple
import PIL
import mmcv
import numpy as np
from detectron2.utils.colormap import colormap
from detectron2.utils.visualizer import VisImage, Visualizer
from mmdet.datasets.coco_panoptic import INSTANCE_OFFSET
from PIL import Image
CLASSES = [
'person', 'bicycle', 'car', 'motorcycle', 'airplane', 'bus', 'train',
'truck', 'boat', 'traffic light', 'fire hydrant', 'stop sign',
'parking meter', 'bench', 'bird', 'cat', 'dog', 'horse', 'sheep', 'cow',
'elephant', 'bear', 'zebra', 'giraffe', 'backpack', 'umbrella', 'handbag',
'tie', 'suitcase', 'frisbee', 'skis', 'snowboard', 'sports ball', 'kite',
'baseball bat', 'baseball glove', 'skateboard', 'surfboard',
'tennis racket', 'bottle', 'wine glass', 'cup', 'fork', 'knife', 'spoon',
'bowl', 'banana', 'apple', 'sandwich', 'orange', 'broccoli', 'carrot',
'hot dog', 'pizza', 'donut', 'cake', 'chair', 'couch', 'potted plant',
'bed', 'dining table', 'toilet', 'tv', 'laptop', 'mouse', 'remote',
'keyboard', 'cell phone', 'microwave', 'oven', 'toaster', 'sink',
'refrigerator', 'book', 'clock', 'vase', 'scissors', 'teddy bear',
'hair drier', 'toothbrush', 'banner', 'blanket', 'bridge', 'cardboard',
'counter', 'curtain', 'door-stuff', 'floor-wood', 'flower', 'fruit',
'gravel', 'house', 'light', 'mirror-stuff', 'net', 'pillow', 'platform',
'playingfield', 'railroad', 'river', 'road', 'roof', 'sand', 'sea',
'shelf', 'snow', 'stairs', 'tent', 'towel', 'wall-brick', 'wall-stone',
'wall-tile', 'wall-wood', 'water-other', 'window-blind', 'window-other',
'tree-merged', 'fence-merged', 'ceiling-merged', 'sky-other-merged',
'cabinet-merged', 'table-merged', 'floor-other-merged', 'pavement-merged',
'mountain-merged', 'grass-merged', 'dirt-merged', 'paper-merged',
'food-other-merged', 'building-other-merged', 'rock-merged',
'wall-other-merged', 'rug-merged', 'background'
]
PREDICATES = [
'over',
'in front of',
'beside',
'on',
'in',
'attached to',
'hanging from',
'on back of',
'falling off',
'going down',
'painted on',
'walking on',
'running on',
'crossing',
'standing on',
'lying on',
'sitting on',
'flying over',
'jumping over',
'jumping from',
'wearing',
'holding',
'carrying',
'looking at',
'guiding',
'kissing',
'eating',
'drinking',
'feeding',
'biting',
'catching',
'picking',
'playing with',
'chasing',
'climbing',
'cleaning',
'playing',
'touching',
'pushing',
'pulling',
'opening',
'cooking',
'talking to',
'throwing',
'slicing',
'driving',
'riding',
'parked on',
'driving on',
'about to hit',
'kicking',
'swinging',
'entering',
'exiting',
'enclosing',
'leaning on',
]
def get_colormap(num_colors: int):
return (np.resize(colormap(), (num_colors, 3))).tolist()
def draw_text(
viz_img: VisImage = None,
text: str = None,
x: float = None,
y: float = None,
color: Tuple[float, float, float] = [0, 0, 0],
size: float = 10,
padding: float = 5,
box_color: str = 'black',
font: str = None,
) -> float:
text_obj = viz_img.ax.text(
x,
y,
text,
size=size,
# family="sans-serif",
bbox={
'facecolor': box_color,
'alpha': 0.8,
'pad': padding,
'edgecolor': 'none',
},
verticalalignment='top',
horizontalalignment='left',
color=color,
zorder=10,
rotation=0,
)
viz_img.get_image()
text_dims = text_obj.get_bbox_patch().get_extents()
return text_dims.width
def show_result(img,
result,
is_one_stage,
num_rel=20,
show=False,
out_dir=None,
out_file=None):
# Load image
img = mmcv.imread(img)
img = img.copy() # (H, W, 3)
img_h, img_w = img.shape[:-1]
# Decrease contrast
img = PIL.Image.fromarray(img)
converter = PIL.ImageEnhance.Color(img)
img = converter.enhance(0.01)
if out_file is not None:
mmcv.imwrite(np.asarray(img), 'bw'+out_file)
# Draw masks
pan_results = result.pan_results
ids = np.unique(pan_results)[::-1]
num_classes = 133
legal_indices = (ids != num_classes) # for VOID label
ids = ids[legal_indices]
# Get predicted labels
labels = np.array([id % INSTANCE_OFFSET for id in ids], dtype=np.int64)
labels = [CLASSES[l] for l in labels]
#For psgtr
rel_obj_labels = result.labels
rel_obj_labels = [CLASSES[l - 1] for l in rel_obj_labels]
# (N_m, H, W)
segms = pan_results[None] == ids[:, None, None]
# Resize predicted masks
segms = [
mmcv.image.imresize(m.astype(float), (img_w, img_h)) for m in segms
]
# One stage segmentation
masks = result.masks
# Choose colors for each instance in coco
colormap_coco = get_colormap(len(masks)) if is_one_stage else get_colormap(len(segms))
colormap_coco = (np.array(colormap_coco) / 255).tolist()
# Viualize masks
viz = Visualizer(img)
viz.overlay_instances(
labels=rel_obj_labels if is_one_stage else labels,
masks=masks if is_one_stage else segms,
assigned_colors=colormap_coco,
)
viz_img = viz.get_output().get_image()
if out_file is not None:
mmcv.imwrite(viz_img, out_file)
# Draw relations
# Filter out relations
### Debug: output all relations if not enough
n_rel_topk = min(num_rel, len(result.labels)//2)
# Exclude background class
rel_dists = result.rel_dists[:, 1:]
# rel_dists = result.rel_dists
rel_scores = rel_dists.max(1)
# rel_scores = result.triplet_scores
# Extract relations with top scores
rel_topk_idx = np.argpartition(rel_scores, -n_rel_topk)[-n_rel_topk:]
rel_labels_topk = rel_dists[rel_topk_idx].argmax(1)
rel_pair_idxes_topk = result.rel_pair_idxes[rel_topk_idx]
relations = np.concatenate(
[rel_pair_idxes_topk, rel_labels_topk[..., None]], axis=1)
n_rels = len(relations)
top_padding = 20
bottom_padding = 20
left_padding = 20
text_size = 10
text_padding = 5
text_height = text_size + 2 * text_padding
row_padding = 10
height = (top_padding + bottom_padding + n_rels *
(text_height + row_padding) - row_padding)
width = img_w
curr_x = left_padding
curr_y = top_padding
# # Adjust colormaps
# colormap_coco = [adjust_text_color(c, viz) for c in colormap_coco]
viz_graph = VisImage(np.full((height, width, 3), 255))
all_rel_vis = []
for i, r in enumerate(relations):
s_idx, o_idx, rel_id = r
s_label = rel_obj_labels[s_idx]
o_label = rel_obj_labels[o_idx]
rel_label = PREDICATES[rel_id]
viz = Visualizer(img)
viz.overlay_instances(
labels=[s_label, o_label],
masks=[masks[s_idx], masks[o_idx]],
assigned_colors=[colormap_coco[s_idx], colormap_coco[o_idx]],
)
viz_masked_img = viz.get_output().get_image()
viz_graph = VisImage(np.full((40, width, 3), 255))
curr_x = 2
curr_y = 2
text_size = 25
text_padding = 20
font = 36
text_width = draw_text(
viz_img=viz_graph,
text=s_label,
x=curr_x,
y=curr_y,
color=colormap_coco[s_idx],
size=text_size,
padding=text_padding,
font=font,
)
curr_x += text_width
# Draw relation text
text_width = draw_text(
viz_img=viz_graph,
text=rel_label,
x=curr_x,
y=curr_y,
size=text_size,
padding=text_padding,
box_color='gainsboro',
font=font,
)
curr_x += text_width
# Draw object text
text_width = draw_text(
viz_img=viz_graph,
text=o_label,
x=curr_x,
y=curr_y,
color=colormap_coco[o_idx],
size=text_size,
padding=text_padding,
font=font,
)
output_viz_graph = np.vstack([viz_masked_img, viz_graph.get_image()])
if show:
all_rel_vis.append(output_viz_graph)
return all_rel_vis
def make_gif(np_images):
frames = [Image.fromarray(numpy_image.astype('uint8'), 'RGB') for numpy_image in np_images]
# frames = [Image.open(image) for image in images]
frame_one = frames[0]
file_name = "top_rel.gif"
frame_one.save(file_name, format="GIF", append_images=frames,
save_all=True, duration=1000, loop=0)
return file_name