Spaces:
Sleeping
Sleeping
Upload app.py
Browse files
app.py
ADDED
|
@@ -0,0 +1,612 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import torch
|
| 2 |
+
import torchvision
|
| 3 |
+
from torchvision.models.detection import FasterRCNN_ResNet50_FPN_Weights
|
| 4 |
+
from transformers import DetrImageProcessor, DetrForObjectDetection
|
| 5 |
+
from transformers import AutoImageProcessor, Mask2FormerForUniversalSegmentation
|
| 6 |
+
from PIL import Image
|
| 7 |
+
import numpy as np
|
| 8 |
+
import matplotlib.pyplot as plt
|
| 9 |
+
import matplotlib.patches as patches
|
| 10 |
+
import gradio as gr
|
| 11 |
+
import os
|
| 12 |
+
import io
|
| 13 |
+
import uuid
|
| 14 |
+
|
| 15 |
+
# Load Faster R-CNN model with proper weight assignment
|
| 16 |
+
frcnn_weights = FasterRCNN_ResNet50_FPN_Weights.DEFAULT
|
| 17 |
+
frcnn_model = torchvision.models.detection.fasterrcnn_resnet50_fpn(weights=None, progress=True)
|
| 18 |
+
state_dict = torch.hub.load_state_dict_from_url(frcnn_weights.url, progress=True, map_location=torch.device('cpu'))
|
| 19 |
+
frcnn_model.load_state_dict(state_dict, strict=False)
|
| 20 |
+
frcnn_model.eval()
|
| 21 |
+
|
| 22 |
+
# Load DETR model and processor
|
| 23 |
+
detr_processor = DetrImageProcessor.from_pretrained("facebook/detr-resnet-50")
|
| 24 |
+
detr_model = DetrForObjectDetection.from_pretrained("facebook/detr-resnet-50")
|
| 25 |
+
|
| 26 |
+
# Load Mask R-CNN model
|
| 27 |
+
maskrcnn_model = torchvision.models.detection.maskrcnn_resnet50_fpn(pretrained=True)
|
| 28 |
+
maskrcnn_model.eval()
|
| 29 |
+
|
| 30 |
+
# Load Mask2Former model and processor
|
| 31 |
+
mask2former_processor = AutoImageProcessor.from_pretrained("facebook/mask2former-swin-small-coco-instance")
|
| 32 |
+
mask2former_model = Mask2FormerForUniversalSegmentation.from_pretrained("facebook/mask2former-swin-small-coco-instance")
|
| 33 |
+
mask2former_model.eval()
|
| 34 |
+
|
| 35 |
+
# COCO class names for Faster R-CNN and Mask R-CNN
|
| 36 |
+
COCO_INSTANCE_CATEGORY_NAMES = [
|
| 37 |
+
'__background__', 'person', 'bicycle', 'car', 'motorcycle', 'airplane', 'bus',
|
| 38 |
+
'train', 'truck', 'boat', 'traffic light', 'fire hydrant', 'N/A', 'stop sign',
|
| 39 |
+
'parking meter', 'bench', 'bird', 'cat', 'dog', 'horse', 'sheep', 'cow',
|
| 40 |
+
'elephant', 'bear', 'zebra', 'giraffe', 'N/A', 'backpack', 'umbrella', 'N/A', 'N/A',
|
| 41 |
+
'handbag', 'tie', 'suitcase', 'frisbee', 'skis', 'snowboard', 'sports ball',
|
| 42 |
+
'kite', 'baseball bat', 'baseball glove', 'skateboard', 'surfboard', 'tennis racket',
|
| 43 |
+
'bottle', 'N/A', 'wine glass', 'cup', 'fork', 'knife', 'spoon', 'bowl',
|
| 44 |
+
'banana', 'apple', 'sandwich', 'orange', 'broccoli', 'carrot', 'hot dog', 'pizza',
|
| 45 |
+
'donut', 'cake', 'chair', 'couch', 'potted plant', 'bed', 'N/A', 'dining table',
|
| 46 |
+
'N/A', 'N/A', 'toilet', 'N/A', 'tv', 'laptop', 'mouse', 'remote', 'keyboard', 'cell phone',
|
| 47 |
+
'microwave', 'oven', 'toaster', 'sink', 'refrigerator', 'N/A', 'book',
|
| 48 |
+
'clock', 'vase', 'scissors', 'teddy bear', 'hair drier', 'toothbrush'
|
| 49 |
+
]
|
| 50 |
+
|
| 51 |
+
# Mask2Former label map
|
| 52 |
+
MASK2FORMER_COCO_NAMES = mask2former_model.config.id2label if hasattr(mask2former_model.config, "id2label") else {str(i): str(i) for i in range(133)}
|
| 53 |
+
|
| 54 |
+
def detect_objects_frcnn(image, threshold=0.5):
|
| 55 |
+
"""Run Faster R-CNN detection."""
|
| 56 |
+
if image is None:
|
| 57 |
+
blank_img = Image.new('RGB', (400, 400), color='white')
|
| 58 |
+
plt.figure(figsize=(10, 10))
|
| 59 |
+
plt.imshow(blank_img)
|
| 60 |
+
plt.text(0.5, 0.5, "No image provided", horizontalalignment='center', verticalalignment='center',
|
| 61 |
+
transform=plt.gca().transAxes, fontsize=20)
|
| 62 |
+
plt.axis('off')
|
| 63 |
+
output_path = f"frcnn_blank_output_{uuid.uuid4()}.png"
|
| 64 |
+
plt.savefig(output_path)
|
| 65 |
+
plt.close()
|
| 66 |
+
return output_path, 0
|
| 67 |
+
|
| 68 |
+
try:
|
| 69 |
+
threshold = float(threshold) if threshold is not None else 0.5
|
| 70 |
+
image = image.convert('RGB')
|
| 71 |
+
img_array = np.array(image).astype(np.float32) / 255.0
|
| 72 |
+
transform = frcnn_weights.transforms()
|
| 73 |
+
image_tensor = transform(Image.fromarray((img_array * 255).astype(np.uint8))).unsqueeze(0)
|
| 74 |
+
|
| 75 |
+
with torch.no_grad():
|
| 76 |
+
prediction = frcnn_model(image_tensor)[0]
|
| 77 |
+
|
| 78 |
+
boxes = prediction['boxes'].cpu().numpy()
|
| 79 |
+
labels = prediction['labels'].cpu().numpy()
|
| 80 |
+
scores = prediction['scores'].cpu().numpy()
|
| 81 |
+
|
| 82 |
+
valid_detections = sum(1 for score in scores if score >= threshold)
|
| 83 |
+
|
| 84 |
+
image_np = np.array(image)
|
| 85 |
+
plt.figure(figsize=(10, 10))
|
| 86 |
+
plt.imshow(image_np)
|
| 87 |
+
ax = plt.gca()
|
| 88 |
+
|
| 89 |
+
for box, label, score in zip(boxes, labels, scores):
|
| 90 |
+
if score >= threshold:
|
| 91 |
+
x1, y1, x2, y2 = box
|
| 92 |
+
ax.add_patch(plt.Rectangle((x1, y1), x2 - x1, y2 - y1, fill=False, color='red', linewidth=2))
|
| 93 |
+
class_name = COCO_INSTANCE_CATEGORY_NAMES[label]
|
| 94 |
+
ax.text(x1, y1, f'{class_name}: {score:.2f}', bbox=dict(facecolor='yellow', alpha=0.5), fontsize=12, color='black')
|
| 95 |
+
|
| 96 |
+
plt.axis('off')
|
| 97 |
+
plt.tight_layout()
|
| 98 |
+
output_path = f"frcnn_output_{uuid.uuid4()}.png"
|
| 99 |
+
plt.savefig(output_path)
|
| 100 |
+
plt.close()
|
| 101 |
+
return output_path, valid_detections
|
| 102 |
+
except Exception as e:
|
| 103 |
+
error_img = Image.new('RGB', (400, 400), color='white')
|
| 104 |
+
plt.figure(figsize=(10, 10))
|
| 105 |
+
plt.imshow(error_img)
|
| 106 |
+
plt.text(0.5, 0.5, f"Error: {str(e)}", horizontalalignment='center', verticalalignment='center',
|
| 107 |
+
transform=plt.gca().transAxes, fontsize=12, wrap=True)
|
| 108 |
+
plt.axis('off')
|
| 109 |
+
error_path = f"frcnn_error_output_{uuid.uuid4()}.png"
|
| 110 |
+
plt.savefig(error_path)
|
| 111 |
+
plt.close()
|
| 112 |
+
return error_path, 0
|
| 113 |
+
|
| 114 |
+
def detect_objects_detr(image, threshold=0.9):
|
| 115 |
+
"""Run DETR detection."""
|
| 116 |
+
if image is None:
|
| 117 |
+
blank_img = Image.new('RGB', (400, 400), color='white')
|
| 118 |
+
fig, ax = plt.subplots(1, figsize=(10, 10))
|
| 119 |
+
ax.imshow(blank_img)
|
| 120 |
+
ax.text(0.5, 0.5, "No image provided", horizontalalignment='center', verticalalignment='center',
|
| 121 |
+
transform=ax.transAxes, fontsize=20)
|
| 122 |
+
plt.axis('off')
|
| 123 |
+
output_path = f"detr_blank_output_{uuid.uuid4()}.png"
|
| 124 |
+
plt.savefig(output_path)
|
| 125 |
+
plt.close(fig)
|
| 126 |
+
return output_path, 0
|
| 127 |
+
|
| 128 |
+
try:
|
| 129 |
+
image = image.convert('RGB')
|
| 130 |
+
inputs = detr_processor(images=image, return_tensors="pt")
|
| 131 |
+
outputs = detr_model(**inputs)
|
| 132 |
+
target_sizes = torch.tensor([image.size[::-1]])
|
| 133 |
+
results = detr_processor.post_process_object_detection(outputs, target_sizes=target_sizes, threshold=threshold)[0]
|
| 134 |
+
|
| 135 |
+
valid_detections = len(results["scores"])
|
| 136 |
+
|
| 137 |
+
fig, ax = plt.subplots(1, figsize=(10, 10))
|
| 138 |
+
ax.imshow(image)
|
| 139 |
+
|
| 140 |
+
for score, label, box in zip(results["scores"], results["labels"], results["boxes"]):
|
| 141 |
+
xmin, ymin, xmax, ymax = box.tolist()
|
| 142 |
+
ax.add_patch(patches.Rectangle((xmin, ymin), xmax - xmin, ymax - ymin, linewidth=2, edgecolor='red', facecolor='none'))
|
| 143 |
+
ax.text(xmin, ymin, f"{detr_model.config.id2label[label.item()]}: {round(score.item(), 2)}",
|
| 144 |
+
bbox=dict(facecolor='yellow', alpha=0.5), fontsize=8)
|
| 145 |
+
|
| 146 |
+
plt.axis('off')
|
| 147 |
+
output_path = f"detr_output_{uuid.uuid4()}.png"
|
| 148 |
+
plt.savefig(output_path)
|
| 149 |
+
plt.close(fig)
|
| 150 |
+
return output_path, valid_detections
|
| 151 |
+
except Exception as e:
|
| 152 |
+
error_img = Image.new('RGB', (400, 400), color='white')
|
| 153 |
+
fig, ax = plt.subplots(1, figsize=(10, 10))
|
| 154 |
+
ax.imshow(error_img)
|
| 155 |
+
ax.text(0.5, 0.5, f"Error: {str(e)}", horizontalalignment='center', verticalalignment='center',
|
| 156 |
+
transform=ax.transAxes, fontsize=12, wrap=True)
|
| 157 |
+
plt.axis('off')
|
| 158 |
+
error_path = f"detr_error_output_{uuid.uuid4()}.png"
|
| 159 |
+
plt.savefig(error_path)
|
| 160 |
+
plt.close(fig)
|
| 161 |
+
return error_path, 0
|
| 162 |
+
|
| 163 |
+
def detect_objects_maskrcnn(image, threshold=0.5):
|
| 164 |
+
"""Run Mask R-CNN detection and segmentation."""
|
| 165 |
+
if image is None:
|
| 166 |
+
blank_img = Image.new('RGB', (400, 400), color='white')
|
| 167 |
+
plt.figure(figsize=(10, 10))
|
| 168 |
+
plt.imshow(blank_img)
|
| 169 |
+
plt.text(0.5, 0.5, "No image provided", horizontalalignment='center', verticalalignment='center',
|
| 170 |
+
transform=plt.gca().transAxes, fontsize=20)
|
| 171 |
+
plt.axis('off')
|
| 172 |
+
output_path = f"maskrcnn_blank_output_{uuid.uuid4()}.png"
|
| 173 |
+
plt.savefig(output_path)
|
| 174 |
+
plt.close()
|
| 175 |
+
return output_path, 0
|
| 176 |
+
|
| 177 |
+
try:
|
| 178 |
+
image = image.convert('RGB')
|
| 179 |
+
transform = torchvision.transforms.ToTensor()
|
| 180 |
+
img_tensor = transform(image).unsqueeze(0)
|
| 181 |
+
|
| 182 |
+
with torch.no_grad():
|
| 183 |
+
output = maskrcnn_model(img_tensor)[0]
|
| 184 |
+
|
| 185 |
+
masks = output['masks']
|
| 186 |
+
boxes = output['boxes'].cpu().numpy()
|
| 187 |
+
labels = output['labels'].cpu().numpy()
|
| 188 |
+
scores = output['scores'].cpu().numpy()
|
| 189 |
+
|
| 190 |
+
valid_detections = sum(1 for score in scores if score >= threshold)
|
| 191 |
+
|
| 192 |
+
image_np = np.array(image).copy()
|
| 193 |
+
fig, ax = plt.subplots(1, figsize=(10, 10))
|
| 194 |
+
ax.imshow(image_np)
|
| 195 |
+
|
| 196 |
+
for i in range(len(masks)):
|
| 197 |
+
if scores[i] >= threshold:
|
| 198 |
+
mask = masks[i, 0].cpu().numpy()
|
| 199 |
+
mask = mask > 0.5
|
| 200 |
+
color = np.random.rand(3)
|
| 201 |
+
colored_mask = np.zeros_like(image_np, dtype=np.uint8)
|
| 202 |
+
for c in range(3):
|
| 203 |
+
colored_mask[:, :, c] = mask * int(color[c] * 255)
|
| 204 |
+
image_np = np.where(mask[:, :, None], 0.5 * image_np + 0.5 * colored_mask, image_np).astype(np.uint8)
|
| 205 |
+
|
| 206 |
+
x1, y1, x2, y2 = boxes[i]
|
| 207 |
+
ax.add_patch(plt.Rectangle((x1, y1), x2 - x1, y2 - y1, fill=False, color=color, linewidth=2))
|
| 208 |
+
label = COCO_INSTANCE_CATEGORY_NAMES[labels[i]]
|
| 209 |
+
ax.text(x1, y1, f"{label}: {scores[i]:.2f}", bbox=dict(facecolor='yellow', alpha=0.5), fontsize=10)
|
| 210 |
+
|
| 211 |
+
ax.imshow(image_np)
|
| 212 |
+
ax.axis('off')
|
| 213 |
+
output_path = f"maskrcnn_output_{uuid.uuid4()}.png"
|
| 214 |
+
plt.savefig(output_path, bbox_inches='tight', pad_inches=0)
|
| 215 |
+
plt.close()
|
| 216 |
+
return output_path, valid_detections
|
| 217 |
+
except Exception as e:
|
| 218 |
+
error_img = Image.new('RGB', (400, 400), color='white')
|
| 219 |
+
plt.figure(figsize=(10, 10))
|
| 220 |
+
plt.imshow(error_img)
|
| 221 |
+
plt.text(0.5, 0.5, f"Error: {str(e)}", horizontalalignment='center', verticalalignment='center',
|
| 222 |
+
transform=plt.gca().transAxes, fontsize=12, wrap=True)
|
| 223 |
+
plt.axis('off')
|
| 224 |
+
error_path = f"maskrcnn_error_output_{uuid.uuid4()}.png"
|
| 225 |
+
plt.savefig(error_path)
|
| 226 |
+
plt.close()
|
| 227 |
+
return error_path, 0
|
| 228 |
+
|
| 229 |
+
def detect_objects_mask2former(image, threshold=0.5):
|
| 230 |
+
"""Run Mask2Former detection and segmentation."""
|
| 231 |
+
if image is None:
|
| 232 |
+
blank_img = Image.new('RGB', (400, 400), color='white')
|
| 233 |
+
plt.figure(figsize=(10, 10))
|
| 234 |
+
plt.imshow(blank_img)
|
| 235 |
+
plt.text(0.5, 0.5, "No image provided", horizontalalignment='center', verticalalignment='center',
|
| 236 |
+
transform=plt.gca().transAxes, fontsize=20)
|
| 237 |
+
plt.axis('off')
|
| 238 |
+
output_path = f"mask2former_blank_output_{uuid.uuid4()}.png"
|
| 239 |
+
plt.savefig(output_path)
|
| 240 |
+
plt.close()
|
| 241 |
+
return output_path, 0
|
| 242 |
+
|
| 243 |
+
try:
|
| 244 |
+
image = image.convert('RGB')
|
| 245 |
+
inputs = mask2former_processor(images=image, return_tensors="pt")
|
| 246 |
+
with torch.no_grad():
|
| 247 |
+
outputs = mask2former_model(**inputs)
|
| 248 |
+
|
| 249 |
+
results = mask2former_processor.post_process_instance_segmentation(outputs, target_sizes=[image.size[::-1]])[0]
|
| 250 |
+
segmentation_map = results["segmentation"].cpu().numpy()
|
| 251 |
+
segments_info = results["segments_info"]
|
| 252 |
+
|
| 253 |
+
valid_detections = sum(1 for segment in segments_info if segment.get("score", 1.0) >= threshold)
|
| 254 |
+
|
| 255 |
+
image_np = np.array(image).copy()
|
| 256 |
+
overlay = image_np.copy()
|
| 257 |
+
fig, ax = plt.subplots(1, figsize=(10, 10))
|
| 258 |
+
ax.imshow(image_np)
|
| 259 |
+
|
| 260 |
+
for segment in segments_info:
|
| 261 |
+
score = segment.get("score", 1.0)
|
| 262 |
+
if score < threshold:
|
| 263 |
+
continue
|
| 264 |
+
segment_id = segment["id"]
|
| 265 |
+
label_id = segment["label_id"]
|
| 266 |
+
mask = segmentation_map == segment_id
|
| 267 |
+
color = np.random.rand(3)
|
| 268 |
+
overlay[mask] = (overlay[mask] * 0.5 + np.array(color) * 255 * 0.5).astype(np.uint8)
|
| 269 |
+
|
| 270 |
+
y_indices, x_indices = np.where(mask)
|
| 271 |
+
if len(x_indices) == 0 or len(y_indices) == 0:
|
| 272 |
+
continue
|
| 273 |
+
x1, x2 = x_indices.min(), x_indices.max()
|
| 274 |
+
y1, y2 = y_indices.min(), y_indices.max()
|
| 275 |
+
|
| 276 |
+
label_name = MASK2FORMER_COCO_NAMES.get(str(label_id), str(label_id))
|
| 277 |
+
ax.add_patch(plt.Rectangle((x1, y1), x2 - x1, y2 - y1, fill=False, color=color, linewidth=2))
|
| 278 |
+
ax.text(x1, y1, f"{label_name}: {score:.2f}", bbox=dict(facecolor='yellow', alpha=0.5), fontsize=10)
|
| 279 |
+
|
| 280 |
+
ax.imshow(overlay)
|
| 281 |
+
ax.axis('off')
|
| 282 |
+
output_path = f"mask2former_output_{uuid.uuid4()}.png"
|
| 283 |
+
plt.savefig(output_path, bbox_inches='tight', pad_inches=0)
|
| 284 |
+
plt.close()
|
| 285 |
+
return output_path, valid_detections
|
| 286 |
+
except Exception as e:
|
| 287 |
+
error_img = Image.new('RGB', (400, 400), color='white')
|
| 288 |
+
plt.figure(figsize=(10, 10))
|
| 289 |
+
plt.imshow(error_img)
|
| 290 |
+
plt.text(0.5, 0.5, f"Error: {str(e)}", horizontalalignment='center', verticalalignment='center',
|
| 291 |
+
transform=plt.gca().transAxes, fontsize=12, wrap=True)
|
| 292 |
+
plt.axis('off')
|
| 293 |
+
error_path = f"mask2former_error_output_{uuid.uuid4()}.png"
|
| 294 |
+
plt.savefig(error_path)
|
| 295 |
+
plt.close()
|
| 296 |
+
return error_path, 0
|
| 297 |
+
|
| 298 |
+
def update_model_choices(category):
|
| 299 |
+
"""Update model choices for prediction radio buttons based on selected category."""
|
| 300 |
+
if category == "Object Detection":
|
| 301 |
+
return gr.update(choices=["ConvNet (Faster R-CNN)", "Transformer (DETR)"], value=None, visible=True), gr.update(visible=True), gr.update(visible=True), gr.update(visible=False), gr.update(visible=False)
|
| 302 |
+
elif category == "Object Segmentation":
|
| 303 |
+
return gr.update(choices=["ConvNet (Mask R-CNN)", "Transformer (Mask2Former)"], value=None, visible=True), gr.update(visible=False), gr.update(visible=False), gr.update(visible=True), gr.update(visible=True)
|
| 304 |
+
return gr.update(choices=[], visible=False), gr.update(visible=False), gr.update(visible=False), gr.update(visible=False), gr.update(visible=False)
|
| 305 |
+
|
| 306 |
+
def analyze_performance(image, category, user_opinion, frcnn_threshold=0.5, detr_threshold=0.9, maskrcnn_threshold=0.5, mask2former_threshold=0.5):
|
| 307 |
+
"""Analyze and compare model performance for all models in the selected category."""
|
| 308 |
+
if image is None:
|
| 309 |
+
return "Please upload an image first.", None, None, None, None, "No analysis available."
|
| 310 |
+
|
| 311 |
+
frcnn_result = None
|
| 312 |
+
detr_result = None
|
| 313 |
+
maskrcnn_result = None
|
| 314 |
+
mask2former_result = None
|
| 315 |
+
frcnn_count = 0
|
| 316 |
+
detr_count = 0
|
| 317 |
+
maskrcnn_count = 0
|
| 318 |
+
mask2former_count = 0
|
| 319 |
+
|
| 320 |
+
if category == "Object Detection":
|
| 321 |
+
frcnn_result, frcnn_count = detect_objects_frcnn(image, frcnn_threshold)
|
| 322 |
+
detr_result, detr_count = detect_objects_detr(image, detr_threshold)
|
| 323 |
+
elif category == "Object Segmentation":
|
| 324 |
+
maskrcnn_result, maskrcnn_count = detect_objects_maskrcnn(image, maskrcnn_threshold)
|
| 325 |
+
mask2former_result, mask2former_count = detect_objects_mask2former(image, mask2former_threshold)
|
| 326 |
+
|
| 327 |
+
# Analyze performance
|
| 328 |
+
counts = {}
|
| 329 |
+
model_mapping = {
|
| 330 |
+
"ConvNet (Faster R-CNN)": "ConvNet (Faster R-CNN)",
|
| 331 |
+
"Transformer (DETR)": "Transformer (DETR)",
|
| 332 |
+
"ConvNet (Mask R-CNN)": "ConvNet (Mask R-CNN)",
|
| 333 |
+
"Transformer (Mask2Former)": "Transformer (Mask2Former)"
|
| 334 |
+
}
|
| 335 |
+
if category == "Object Detection":
|
| 336 |
+
counts = {
|
| 337 |
+
"ConvNet (Faster R-CNN)": frcnn_count,
|
| 338 |
+
"Transformer (DETR)": detr_count
|
| 339 |
+
}
|
| 340 |
+
elif category == "Object Segmentation":
|
| 341 |
+
counts = {
|
| 342 |
+
"ConvNet (Mask R-CNN)": maskrcnn_count,
|
| 343 |
+
"Transformer (Mask2Former)": mask2former_count
|
| 344 |
+
}
|
| 345 |
+
|
| 346 |
+
max_count = max(counts.values())
|
| 347 |
+
max_models = [model for model, count in counts.items() if count == max_count]
|
| 348 |
+
|
| 349 |
+
if len(max_models) == 1:
|
| 350 |
+
analysis = f"Result: {max_models[0]} performed best, identifying {max_count} objects.\n\n"
|
| 351 |
+
else:
|
| 352 |
+
analysis = f"Result: {', '.join(max_models)} performed equally well, each identifying {max_count} objects.\n\n"
|
| 353 |
+
|
| 354 |
+
if user_opinion:
|
| 355 |
+
analysis += f"You predicted that {user_opinion} would perform best.\n"
|
| 356 |
+
if user_opinion in max_models:
|
| 357 |
+
analysis += f"Congratulations, your prediction was correct!\n"
|
| 358 |
+
else:
|
| 359 |
+
analysis += f"Your prediction was not correct. {user_opinion} identified {counts[user_opinion]} objects, while {', '.join(max_models)} performed best with {max_count} objects. Please try again with a new image.\n"
|
| 360 |
+
|
| 361 |
+
if category == "Object Detection":
|
| 362 |
+
analysis += "\nConvNet (Faster R-CNN) is efficient and reliable for general object identification tasks. Transformer (DETR) excels in complex scenes by leveraging advanced context understanding."
|
| 363 |
+
elif category == "Object Segmentation":
|
| 364 |
+
analysis += "\nConvNet (Mask R-CNN) provides precise object outlines for detailed analysis. Transformer (Mask2Former) often outperforms in complex scenes due to its advanced architecture."
|
| 365 |
+
|
| 366 |
+
# Image-specific recommendation
|
| 367 |
+
img_array = np.array(image)
|
| 368 |
+
height, width = img_array.shape[:2]
|
| 369 |
+
pixel_variance = np.var(img_array)
|
| 370 |
+
|
| 371 |
+
if height * width > 1000 * 1000:
|
| 372 |
+
analysis += f"\n\nThis high-resolution image benefits from Transformer models, which excel in detailed and complex scenes."
|
| 373 |
+
if pixel_variance > 1000:
|
| 374 |
+
analysis += f"\n\nThis image has high complexity. Transformer models often provide superior results in such cases."
|
| 375 |
+
if height * width < 500 * 500:
|
| 376 |
+
analysis += f"\n\nFor smaller images, ConvNet models often deliver reliable results with lower computational demands."
|
| 377 |
+
if category == "Object Segmentation" and max_count > 0:
|
| 378 |
+
analysis += "\n\nFor detailed outlining tasks, Transformer (Mask2Former) may be preferable for complex scenes due to its advanced design."
|
| 379 |
+
|
| 380 |
+
# Enhanced result formatting
|
| 381 |
+
if user_opinion and user_opinion in max_models:
|
| 382 |
+
celebration = "๐โจ"
|
| 383 |
+
analysis = analysis.replace("Congratulations", f"{celebration} EPIC WIN! {celebration}")
|
| 384 |
+
analysis = analysis.replace("!\n", "! ๐ฅณ\n")
|
| 385 |
+
analysis += "\n\n๐ You've mastered the AI showdown! ๐"
|
| 386 |
+
elif user_opinion:
|
| 387 |
+
analysis = analysis.replace("try again", "try again ๐ช")
|
| 388 |
+
|
| 389 |
+
# Convert to HTML with styling
|
| 390 |
+
html_analysis = f"""
|
| 391 |
+
<div class="{'celebrate' if user_opinion in max_models else ''}" style="margin: 15px 0;">
|
| 392 |
+
<h3 style='color: {"#4CAF50" if user_opinion in max_models else "#f44336"}; margin-bottom: 15px;'>
|
| 393 |
+
{"๐ " + max_models[0] + " Dominates!" if len(max_models) == 1 else "โ๏ธ Tie Battle!"}
|
| 394 |
+
</h3>
|
| 395 |
+
<div style="background: var(--background-fill-primary); padding: 20px; border-radius: 10px;
|
| 396 |
+
white-space: pre-wrap; overflow-wrap: break-word; color: var(--text-color);">
|
| 397 |
+
{analysis}
|
| 398 |
+
</div>
|
| 399 |
+
</div>
|
| 400 |
+
"""
|
| 401 |
+
return "Analysis complete!", frcnn_result, detr_result, maskrcnn_result, mask2former_result, html_analysis
|
| 402 |
+
|
| 403 |
+
# Create Gradio interface with enhanced design
|
| 404 |
+
with gr.Blocks(title="AI Vision Showdown", theme=gr.themes.Default(primary_hue="emerald", secondary_hue="blue")) as app:
|
| 405 |
+
gr.Markdown("""
|
| 406 |
+
# ๐ฏ AI Vision Showdown: ConvNets vs Transformers
|
| 407 |
+
### ๐ค Battle of the algorithms! Upload an image and predict which AI will dominate!
|
| 408 |
+
""")
|
| 409 |
+
|
| 410 |
+
# Enhanced CSS
|
| 411 |
+
gr.HTML("""
|
| 412 |
+
<style>
|
| 413 |
+
@keyframes celebrate {
|
| 414 |
+
0% { transform: rotate(0deg); }
|
| 415 |
+
25% { transform: rotate(5deg); }
|
| 416 |
+
50% { transform: rotate(-5deg); }
|
| 417 |
+
75% { transform: rotate(5deg); }
|
| 418 |
+
100% { transform: rotate(0deg); }
|
| 419 |
+
}
|
| 420 |
+
.celebrate { animation: celebrate 0.5s ease-in-out; }
|
| 421 |
+
.battle-card {
|
| 422 |
+
border-radius: 15px;
|
| 423 |
+
padding: 20px;
|
| 424 |
+
margin: 10px 0;
|
| 425 |
+
background: var(--background-fill-primary);
|
| 426 |
+
border: 1px solid var(--border-color-primary);
|
| 427 |
+
}
|
| 428 |
+
.analysis-box {
|
| 429 |
+
background: var(--background-fill-secondary) !important;
|
| 430 |
+
color: var(--text-color) !important;
|
| 431 |
+
padding: 20px;
|
| 432 |
+
border-radius: 10px;
|
| 433 |
+
white-space: pre-wrap;
|
| 434 |
+
overflow-wrap: break-word;
|
| 435 |
+
}
|
| 436 |
+
.loading-status {
|
| 437 |
+
padding: 15px;
|
| 438 |
+
background: var(--background-fill-secondary);
|
| 439 |
+
border-radius: 8px;
|
| 440 |
+
margin: 10px 0;
|
| 441 |
+
text-align: center;
|
| 442 |
+
font-weight: bold;
|
| 443 |
+
}
|
| 444 |
+
</style>
|
| 445 |
+
""")
|
| 446 |
+
|
| 447 |
+
# State variables
|
| 448 |
+
image_state = gr.State(None)
|
| 449 |
+
category_state = gr.State(None)
|
| 450 |
+
loading_status = gr.HTML(visible=False)
|
| 451 |
+
|
| 452 |
+
# Top Section: Inputs
|
| 453 |
+
with gr.Row(variant="battle-card"):
|
| 454 |
+
with gr.Column(scale=1, min_width=300):
|
| 455 |
+
gr.Markdown("## ๐ค Image Upload Zone")
|
| 456 |
+
image_input = gr.Image(type="pil", label="Drag & Drop Your Challenge Image")
|
| 457 |
+
upload_button = gr.Button("๐ผ Upload Challenge Image", variant="primary")
|
| 458 |
+
|
| 459 |
+
with gr.Column(scale=1, min_width=300):
|
| 460 |
+
with gr.Group(visible=False) as prediction_selection:
|
| 461 |
+
gr.Markdown("## ๐ฎ Prediction Arena")
|
| 462 |
+
category_choice = gr.Radio(
|
| 463 |
+
choices=["Object Detection", "Object Segmentation"],
|
| 464 |
+
label="โ๏ธ Select Battle Ground",
|
| 465 |
+
value=None,
|
| 466 |
+
elem_classes="battle-card"
|
| 467 |
+
)
|
| 468 |
+
user_opinion = gr.Radio(
|
| 469 |
+
choices=[],
|
| 470 |
+
label="๐น Predict the Victor",
|
| 471 |
+
value=None,
|
| 472 |
+
visible=False,
|
| 473 |
+
elem_classes="battle-card"
|
| 474 |
+
)
|
| 475 |
+
|
| 476 |
+
# Enhanced threshold controls
|
| 477 |
+
with gr.Accordion("๐๏ธ Advanced Battle Parameters", open=False):
|
| 478 |
+
frcnn_threshold = gr.Slider(
|
| 479 |
+
minimum=0.0, maximum=1.0, value=0.5, step=0.05,
|
| 480 |
+
label="Faster R-CNN Confidence (Speed Demon ๐๏ธ)",
|
| 481 |
+
visible=False
|
| 482 |
+
)
|
| 483 |
+
detr_threshold = gr.Slider(
|
| 484 |
+
minimum=0.0, maximum=1.0, value=0.9, step=0.05,
|
| 485 |
+
label="DETR Confidence (Attention Master ๐)",
|
| 486 |
+
visible=False
|
| 487 |
+
)
|
| 488 |
+
maskrcnn_threshold = gr.Slider(
|
| 489 |
+
minimum=0.0, maximum=1.0, value=0.5, step=0.05,
|
| 490 |
+
label="Mask R-CNN Confidence (Precision Expert โ๏ธ)",
|
| 491 |
+
visible=False
|
| 492 |
+
)
|
| 493 |
+
mask2former_threshold = gr.Slider(
|
| 494 |
+
minimum=0.0, maximum=1.0, value=0.5, step=0.05,
|
| 495 |
+
label="Mask2Former Confidence (Transformer Champ ๐ค)",
|
| 496 |
+
visible=False
|
| 497 |
+
)
|
| 498 |
+
|
| 499 |
+
detect_button = gr.Button("โ๏ธ Start Showdown", variant="primary")
|
| 500 |
+
|
| 501 |
+
# Results Section
|
| 502 |
+
with gr.Group(visible=False) as outputs_panel:
|
| 503 |
+
gr.Markdown("## ๐ Battle Results")
|
| 504 |
+
with gr.Tabs():
|
| 505 |
+
with gr.TabItem("Object Detection Warriors", visible=False) as detection_tab:
|
| 506 |
+
with gr.Row():
|
| 507 |
+
frcnn_result = gr.Image(type="filepath", label="๐ Faster R-CNN (ConvNet Champion)", elem_classes="battle-card")
|
| 508 |
+
detr_result = gr.Image(type="filepath", label="๐ง DETR (Transformer Visionary)", elem_classes="battle-card")
|
| 509 |
+
|
| 510 |
+
with gr.TabItem("Segmentation Gladiators", visible=False) as segmentation_tab:
|
| 511 |
+
with gr.Row():
|
| 512 |
+
maskrcnn_result = gr.Image(type="filepath", label="โ๏ธ Mask R-CNN (Pixel Perfect)", elem_classes="battle-card")
|
| 513 |
+
mask2former_result = gr.Image(type="filepath", label="๐ก๏ธ Mask2Former (Segmentation Master)", elem_classes="battle-card")
|
| 514 |
+
|
| 515 |
+
# Analysis Section
|
| 516 |
+
with gr.Group(visible=False) as results_panel:
|
| 517 |
+
gr.Markdown("## ๐ Battle Report")
|
| 518 |
+
analysis_output = gr.HTML(label="Victory Analysis", elem_classes="battle-card")
|
| 519 |
+
restart_button = gr.Button("๐ New Challenge", variant="secondary")
|
| 520 |
+
|
| 521 |
+
# Upload button click event
|
| 522 |
+
def upload_image(img):
|
| 523 |
+
if img is None:
|
| 524 |
+
return None, gr.update(visible=False), gr.update(visible=False), gr.update(visible=False)
|
| 525 |
+
return img, gr.update(visible=True), gr.update(visible=False), gr.update(visible=False)
|
| 526 |
+
|
| 527 |
+
upload_button.click(
|
| 528 |
+
fn=upload_image,
|
| 529 |
+
inputs=[image_input],
|
| 530 |
+
outputs=[image_state, prediction_selection, outputs_panel, results_panel]
|
| 531 |
+
)
|
| 532 |
+
|
| 533 |
+
# Category selection event
|
| 534 |
+
def update_prediction_options(category):
|
| 535 |
+
if category is None:
|
| 536 |
+
return None, gr.update(visible=False), gr.update(visible=False), gr.update(visible=False), gr.update(visible=False), gr.update(visible=False)
|
| 537 |
+
model_update, frcnn_vis, detr_vis, maskrcnn_vis, mask2former_vis = update_model_choices(category)
|
| 538 |
+
return category, model_update, frcnn_vis, detr_vis, maskrcnn_vis, mask2former_vis
|
| 539 |
+
|
| 540 |
+
category_choice.change(
|
| 541 |
+
fn=update_prediction_options,
|
| 542 |
+
inputs=[category_choice],
|
| 543 |
+
outputs=[category_state, user_opinion, frcnn_threshold, detr_threshold, maskrcnn_threshold, mask2former_threshold]
|
| 544 |
+
)
|
| 545 |
+
|
| 546 |
+
# Detect button click event
|
| 547 |
+
def run_detection(image, category, user_opinion, frcnn_threshold, detr_threshold, maskrcnn_threshold, mask2former_threshold):
|
| 548 |
+
if not category or not user_opinion:
|
| 549 |
+
return "Please select a category and prediction.", None, None, None, None, "No analysis available.", gr.update(visible=False), gr.update(visible=False), gr.update(visible=False), gr.update(visible=False)
|
| 550 |
+
|
| 551 |
+
def analyze_with_progress(progress=gr.Progress()):
|
| 552 |
+
progress(0.1, desc="โ๏ธ Models are gearing up...")
|
| 553 |
+
result = analyze_performance(image, category, user_opinion, frcnn_threshold, detr_threshold, maskrcnn_threshold, mask2former_threshold)
|
| 554 |
+
progress(1.0, desc="โ
Battle complete!")
|
| 555 |
+
return result
|
| 556 |
+
|
| 557 |
+
try:
|
| 558 |
+
message, frcnn_result_img, detr_result_img, maskrcnn_result_img, mask2former_result_img, html_analysis = analyze_with_progress()
|
| 559 |
+
return [
|
| 560 |
+
message,
|
| 561 |
+
gr.update(value=frcnn_result_img, visible=category == "Object Detection"),
|
| 562 |
+
gr.update(value=detr_result_img, visible=category == "Object Detection"),
|
| 563 |
+
gr.update(value=maskrcnn_result_img, visible=category == "Object Segmentation"),
|
| 564 |
+
gr.update(value=mask2former_result_img, visible=category == "Object Segmentation"),
|
| 565 |
+
html_analysis,
|
| 566 |
+
gr.update(visible=True),
|
| 567 |
+
gr.update(visible=True),
|
| 568 |
+
gr.update(visible=category == "Object Detection"),
|
| 569 |
+
gr.update(visible=category == "Object Segmentation"),
|
| 570 |
+
gr.update(visible=False)
|
| 571 |
+
]
|
| 572 |
+
except Exception as e:
|
| 573 |
+
return [f"Error: {str(e)}"] + [gr.update()]*9 + [gr.update(visible=False)]
|
| 574 |
+
|
| 575 |
+
detect_button.click(
|
| 576 |
+
fn=run_detection,
|
| 577 |
+
inputs=[image_state, category_state, user_opinion, frcnn_threshold, detr_threshold, maskrcnn_threshold, mask2former_threshold],
|
| 578 |
+
outputs=[gr.Textbox(visible=False), frcnn_result, detr_result, maskrcnn_result, mask2former_result,
|
| 579 |
+
analysis_output, outputs_panel, results_panel, detection_tab, segmentation_tab, loading_status]
|
| 580 |
+
)
|
| 581 |
+
|
| 582 |
+
# Restart button click event
|
| 583 |
+
def restart():
|
| 584 |
+
return None, None, gr.update(visible=False), gr.update(visible=False), gr.update(visible=False), gr.update(visible=False), gr.update(visible=False), gr.update(visible=False), gr.update(visible=False), gr.update(visible=False), gr.update(visible=False), gr.update(visible=False), gr.update(visible=False)
|
| 585 |
+
|
| 586 |
+
restart_button.click(
|
| 587 |
+
fn=restart,
|
| 588 |
+
inputs=[],
|
| 589 |
+
outputs=[image_state, category_state, prediction_selection, outputs_panel, results_panel, frcnn_result, detr_result, maskrcnn_result, mask2former_result, analysis_output, user_opinion, category_choice, detection_tab, segmentation_tab]
|
| 590 |
+
)
|
| 591 |
+
|
| 592 |
+
# Example images
|
| 593 |
+
example_images = [
|
| 594 |
+
os.path.join(os.getcwd(), "TEST_IMG_1.jpg"),
|
| 595 |
+
os.path.join(os.getcwd(), "TEST_IMG_2.JPG"),
|
| 596 |
+
os.path.join(os.getcwd(), "TEST_IMG_3.jpg"),
|
| 597 |
+
os.path.join(os.getcwd(), "TEST_IMG_4.jpg")
|
| 598 |
+
]
|
| 599 |
+
|
| 600 |
+
valid_examples = [img for img in example_images if os.path.exists(img)]
|
| 601 |
+
|
| 602 |
+
if valid_examples:
|
| 603 |
+
gr.Markdown("## ๐งฉ Try These Example Challenges:")
|
| 604 |
+
gr.Examples(
|
| 605 |
+
examples=valid_examples,
|
| 606 |
+
inputs=image_input,
|
| 607 |
+
examples_per_page=4,
|
| 608 |
+
label=""
|
| 609 |
+
)
|
| 610 |
+
|
| 611 |
+
if __name__ == "__main__":
|
| 612 |
+
app.launch(debug=True)
|