Spaces:
Runtime error
Runtime error
feat: ✨ YOLO-World-Seg Image process added
Browse filesSigned-off-by: Onuralp SEZER <[email protected]>
- README.md +3 -3
- app.py +184 -10
- requirements.txt +12 -12
README.md
CHANGED
@@ -1,13 +1,13 @@
|
|
1 |
---
|
2 |
title: YOLO World Seg
|
3 |
-
emoji:
|
4 |
colorFrom: purple
|
5 |
-
colorTo:
|
6 |
sdk: gradio
|
7 |
sdk_version: 4.19.1
|
8 |
app_file: app.py
|
9 |
pinned: false
|
10 |
-
license:
|
11 |
---
|
12 |
- openai/clip-vit-base-patch32
|
13 |
- wondervictor/YOLO-World
|
|
|
1 |
---
|
2 |
title: YOLO World Seg
|
3 |
+
emoji: 🎨
|
4 |
colorFrom: purple
|
5 |
+
colorTo: red
|
6 |
sdk: gradio
|
7 |
sdk_version: 4.19.1
|
8 |
app_file: app.py
|
9 |
pinned: false
|
10 |
+
license: gpl-3.0
|
11 |
---
|
12 |
- openai/clip-vit-base-patch32
|
13 |
- wondervictor/YOLO-World
|
app.py
CHANGED
@@ -1,16 +1,190 @@
|
|
1 |
-
|
2 |
-
|
3 |
-
|
4 |
-
|
5 |
-
|
6 |
|
7 |
|
8 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
9 |
|
10 |
import gradio as gr
|
11 |
|
12 |
-
def greet(name):
|
13 |
-
return "text"
|
14 |
|
15 |
-
|
16 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import os
|
2 |
+
os.system("mim install 'mmengine>=0.7.0'")
|
3 |
+
os.system("mim install mmcv")
|
4 |
+
os.system("mim install 'mmdet>=3.0.0'")
|
5 |
+
os.system("pip install -e .")
|
6 |
|
7 |
|
8 |
+
import numpy as np
|
9 |
+
import torch
|
10 |
+
from mmengine.config import Config
|
11 |
+
from mmengine.dataset import Compose
|
12 |
+
from mmengine.runner import Runner
|
13 |
+
from mmengine.runner.amp import autocast
|
14 |
+
from mmyolo.registry import RUNNERS
|
15 |
+
from torchvision.ops import nms
|
16 |
+
import supervision as sv
|
17 |
+
import PIL.Image
|
18 |
+
import cv2
|
19 |
|
20 |
import gradio as gr
|
21 |
|
|
|
|
|
22 |
|
23 |
+
TITLE = """
|
24 |
+
# YOLO-World-Seg
|
25 |
+
|
26 |
+
This is a demo of zero-shot object detection and instance segmentation using
|
27 |
+
[YOLO-World](https://github.com/AILab-CVC/YOLO-World)
|
28 |
+
|
29 |
+
Powered by [Supervision](https://github.com/roboflow/supervision).
|
30 |
+
"""
|
31 |
+
|
32 |
+
EXAMPLES = [
|
33 |
+
["https://media.roboflow.com/efficient-sam/corgi.jpg", "dog",0.5,0.5,0.5,100],
|
34 |
+
["https://media.roboflow.com/efficient-sam/horses.jpg", "horse",0.5,0.5,0.5,100],
|
35 |
+
["https://media.roboflow.com/efficient-sam/bears.jpg", "bear",0.5,0.5,0.5,100],
|
36 |
+
]
|
37 |
+
|
38 |
+
box_annotator = sv.BoxAnnotator()
|
39 |
+
label_annotator = sv.LabelAnnotator(text_position=sv.Position.CENTER)
|
40 |
+
mask_annotator = sv.MaskAnnotator(color_lookup=sv.ColorLookup.INDEX)
|
41 |
+
|
42 |
+
def load_runner():
|
43 |
+
cfg = Config.fromfile(
|
44 |
+
"./configs/segmentation/yolo_world_seg_l_dual_vlpan_2e-4_80e_8gpus_seghead_finetune_lvis.py"
|
45 |
+
)
|
46 |
+
cfg.work_dir = "."
|
47 |
+
cfg.load_from = "yolo_world_seg_l_dual_vlpan_2e-4_80e_8gpus_seghead_finetune_lvis-5a642d30.pth"
|
48 |
+
runner = Runner.from_cfg(cfg)
|
49 |
+
runner.call_hook("before_run")
|
50 |
+
runner.load_or_resume()
|
51 |
+
pipeline = cfg.test_dataloader.dataset.pipeline
|
52 |
+
runner.pipeline = Compose(pipeline)
|
53 |
+
runner.model.eval()
|
54 |
+
|
55 |
+
def run_image(
|
56 |
+
input_image,
|
57 |
+
class_names="person,car,bus,truck",
|
58 |
+
score_thr=0.05,
|
59 |
+
iou_thr=0.5,
|
60 |
+
nms_thr = 0.5,
|
61 |
+
max_num_boxes=100,
|
62 |
+
):
|
63 |
+
runner = load_runner()
|
64 |
+
with open("input.jpeg", "wb") as f:
|
65 |
+
f.write(input_image)
|
66 |
+
|
67 |
+
class_names = [class_name.strip() for class_name in class_names.split(',')]
|
68 |
+
|
69 |
+
texts = [[t.strip()] for t in class_names.split(",")] + [[" "]]
|
70 |
+
data_info = runner.pipeline(dict(img_id=0, img_path="input.jpeg",
|
71 |
+
texts=texts))
|
72 |
+
|
73 |
+
data_batch = dict(
|
74 |
+
inputs=data_info["inputs"].unsqueeze(0),
|
75 |
+
data_samples=[data_info["data_samples"]],
|
76 |
+
)
|
77 |
+
|
78 |
+
with autocast(enabled=False), torch.no_grad():
|
79 |
+
output = runner.model.test_step(data_batch)[0]
|
80 |
+
runner.model.class_names = texts
|
81 |
+
pred_instances = output.pred_instances
|
82 |
+
|
83 |
+
keep_idxs = nms(pred_instances.bboxes, pred_instances.scores, iou_threshold=iou_thr)
|
84 |
+
pred_instances = pred_instances[keep_idxs]
|
85 |
+
pred_instances = pred_instances[pred_instances.scores.float() > score_thr]
|
86 |
+
|
87 |
+
if len(pred_instances.scores) > max_num_boxes:
|
88 |
+
indices = pred_instances.scores.float().topk(max_num_boxes)[1]
|
89 |
+
pred_instances = pred_instances[indices]
|
90 |
+
output.pred_instances = pred_instances
|
91 |
+
result = pred_instances.cpu().numpy()
|
92 |
+
detections = sv.Detections(
|
93 |
+
xyxy=result['bboxes'],
|
94 |
+
class_id=result['labels'],
|
95 |
+
confidence=result['scores'],
|
96 |
+
mask = result['masks']
|
97 |
+
)
|
98 |
+
detections = detections.with_nms(threshold=nms_thr)
|
99 |
+
|
100 |
+
labels = [
|
101 |
+
f"{class_id} {confidence:.3f}"
|
102 |
+
for class_id, confidence
|
103 |
+
in zip(detections.class_id, detections.confidence)
|
104 |
+
]
|
105 |
+
|
106 |
+
svimage = box_annotator.annotate(input_image, detections)
|
107 |
+
svimage = label_annotator.annotate(svimage, detections, labels)
|
108 |
+
svimage = mask_annotator.annotate(svimage,detections)
|
109 |
+
return svimage
|
110 |
+
|
111 |
+
confidence_threshold_component = gr.Slider(
|
112 |
+
minimum=0,
|
113 |
+
maximum=1.0,
|
114 |
+
value=0.3,
|
115 |
+
step=0.01,
|
116 |
+
label="Confidence Threshold",
|
117 |
+
info=(
|
118 |
+
"The confidence threshold for the YOLO-World model. Lower the threshold to "
|
119 |
+
"reduce false negatives, enhancing the model's sensitivity to detect "
|
120 |
+
"sought-after objects. Conversely, increase the threshold to minimize false "
|
121 |
+
"positives, preventing the model from identifying objects it shouldn't."
|
122 |
+
))
|
123 |
+
|
124 |
+
iou_threshold_component = gr.Slider(
|
125 |
+
minimum=0,
|
126 |
+
maximum=1.0,
|
127 |
+
value=0.5,
|
128 |
+
step=0.01,
|
129 |
+
label="IoU Threshold",
|
130 |
+
info=(
|
131 |
+
"The Intersection over Union (IoU) threshold for non-maximum suppression. "
|
132 |
+
"Decrease the value to lessen the occurrence of overlapping bounding boxes, "
|
133 |
+
"making the detection process stricter. On the other hand, increase the value "
|
134 |
+
"to allow more overlapping bounding boxes, accommodating a broader range of "
|
135 |
+
"detections."
|
136 |
+
))
|
137 |
+
|
138 |
+
with gr.Blocks() as demo:
|
139 |
+
gr.Markdown(TITLE)
|
140 |
+
with gr.Accordion("Configuration", open=False):
|
141 |
+
confidence_threshold_component.render()
|
142 |
+
iou_threshold_component.render()
|
143 |
+
with gr.Tab(label="Image"):
|
144 |
+
with gr.Row():
|
145 |
+
input_image_component = gr.Image(
|
146 |
+
type='numpy',
|
147 |
+
label='Input Image'
|
148 |
+
)
|
149 |
+
output_image_component = gr.Image(
|
150 |
+
type='numpy',
|
151 |
+
label='Output Image'
|
152 |
+
)
|
153 |
+
with gr.Row():
|
154 |
+
image_categories_text_component = gr.Textbox(
|
155 |
+
label='Categories',
|
156 |
+
placeholder='comma separated list of categories',
|
157 |
+
scale=7
|
158 |
+
)
|
159 |
+
image_submit_button_component = gr.Button(
|
160 |
+
value='Submit',
|
161 |
+
scale=1,
|
162 |
+
variant='primary'
|
163 |
+
)
|
164 |
+
gr.Examples(
|
165 |
+
fn=run_image,
|
166 |
+
examples=EXAMPLES,
|
167 |
+
inputs=[
|
168 |
+
input_image_component,
|
169 |
+
image_categories_text_component,
|
170 |
+
confidence_threshold_component,
|
171 |
+
iou_threshold_component,
|
172 |
+
],
|
173 |
+
outputs=output_image_component
|
174 |
+
)
|
175 |
+
|
176 |
+
|
177 |
+
image_submit_button_component.click(
|
178 |
+
fn=run_image,
|
179 |
+
inputs=[
|
180 |
+
input_image_component,
|
181 |
+
image_categories_text_component,
|
182 |
+
confidence_threshold_component,
|
183 |
+
iou_threshold_component,
|
184 |
+
],
|
185 |
+
outputs=output_image_component
|
186 |
+
)
|
187 |
+
|
188 |
+
|
189 |
+
|
190 |
+
demo.launch(debug=False, show_error=True)
|
requirements.txt
CHANGED
@@ -1,14 +1,14 @@
|
|
1 |
-
|
2 |
gradio
|
3 |
-
|
4 |
-
|
5 |
-
|
6 |
-
|
7 |
-
|
8 |
|
9 |
-
|
10 |
-
|
11 |
-
|
12 |
-
|
13 |
-
|
14 |
-
|
|
|
1 |
+
openmim
|
2 |
gradio
|
3 |
+
transformers
|
4 |
+
numpy
|
5 |
+
opencv-python
|
6 |
+
supervision
|
7 |
+
wheel
|
8 |
|
9 |
+
--extra-index-url https://download.pytorch.org/whl/cu121
|
10 |
+
torch==2.1.0+cu121
|
11 |
+
torchdata==0.7.0
|
12 |
+
torchsummary==1.5.1
|
13 |
+
torchtext==0.16.0
|
14 |
+
torchvision==0.16.0+cu121
|