Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
@@ -140,51 +140,75 @@ def match_and_identify(features, bbox):
|
|
140 |
return identity, color
|
141 |
|
142 |
def process_image(image):
|
143 |
-
|
144 |
-
|
145 |
-
input_tensor = np.expand_dims(image_np, axis=0)
|
146 |
-
|
147 |
-
# Run inference
|
148 |
-
detections = detect_objects(input_tensor)
|
149 |
-
|
150 |
-
# Extract output tensors and convert to numpy arrays
|
151 |
-
boxes = detections[0].numpy()[0]
|
152 |
-
scores = detections[1].numpy()[0]
|
153 |
-
classes = detections[2].numpy()[0]
|
154 |
-
num_detections = int(detections[3].numpy()[0])
|
155 |
-
|
156 |
-
# Filter detections for 'person' class
|
157 |
-
threshold = 0.3 # Adjust this threshold as needed
|
158 |
-
for i in range(num_detections):
|
159 |
-
class_id = int(classes[i])
|
160 |
-
score = scores[i]
|
161 |
-
box = boxes[i]
|
162 |
|
163 |
-
|
164 |
-
|
165 |
-
|
166 |
-
|
167 |
-
|
168 |
-
# Extract person ROI
|
169 |
-
person_roi = image[top:bottom, left:right]
|
170 |
|
171 |
-
|
172 |
-
|
|
|
173 |
|
174 |
-
|
175 |
-
|
176 |
-
|
177 |
-
# Match and identify
|
178 |
-
identity, color = match_and_identify(features, predicted_bbox)
|
179 |
|
180 |
-
|
181 |
-
|
182 |
-
|
183 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
184 |
|
185 |
return image
|
186 |
-
|
187 |
def gradio_interface(input_image):
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
188 |
# Process the input image
|
189 |
output_image = process_image(input_image)
|
190 |
return output_image
|
|
|
140 |
return identity, color
|
141 |
|
142 |
def process_image(image):
|
143 |
+
if image is None:
|
144 |
+
return None
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
145 |
|
146 |
+
# Convert image to RGB if it's not
|
147 |
+
if len(image.shape) == 2: # Grayscale
|
148 |
+
image = cv2.cvtColor(image, cv2.COLOR_GRAY2RGB)
|
149 |
+
elif image.shape[2] == 4: # RGBA
|
150 |
+
image = cv2.cvtColor(image, cv2.COLOR_RGBA2RGB)
|
|
|
|
|
151 |
|
152 |
+
# Ensure image is uint8
|
153 |
+
if image.dtype != np.uint8:
|
154 |
+
image = (image * 255).astype(np.uint8)
|
155 |
|
156 |
+
# Prepare the image tensor
|
157 |
+
image_np = np.array(image)
|
158 |
+
input_tensor = np.expand_dims(image_np, axis=0)
|
|
|
|
|
159 |
|
160 |
+
try:
|
161 |
+
# Run inference
|
162 |
+
detections = detect_objects(input_tensor)
|
163 |
+
|
164 |
+
# Extract output tensors and convert to numpy arrays
|
165 |
+
boxes = detections[0].numpy()[0]
|
166 |
+
scores = detections[1].numpy()[0]
|
167 |
+
classes = detections[2].numpy()[0]
|
168 |
+
num_detections = int(detections[3].numpy()[0])
|
169 |
+
|
170 |
+
# Filter detections for 'person' class
|
171 |
+
threshold = 0.3 # Adjust this threshold as needed
|
172 |
+
for i in range(num_detections):
|
173 |
+
class_id = int(classes[i])
|
174 |
+
score = scores[i]
|
175 |
+
box = boxes[i]
|
176 |
+
|
177 |
+
if class_id == 1 and score > threshold:
|
178 |
+
h, w, _ = image.shape
|
179 |
+
ymin, xmin, ymax, xmax = box
|
180 |
+
left, right, top, bottom = int(xmin * w), int(xmax * w), int(ymin * h), int(ymax * h)
|
181 |
+
|
182 |
+
# Extract person ROI
|
183 |
+
person_roi = image[top:bottom, left:right]
|
184 |
+
|
185 |
+
# Extract features
|
186 |
+
features = extract_features(person_roi)
|
187 |
+
|
188 |
+
# Predict bbox using Kalman filter
|
189 |
+
predicted_bbox = np.array([xmin, ymin, xmax, ymax])
|
190 |
+
|
191 |
+
# Match and identify
|
192 |
+
identity, color = match_and_identify(features, predicted_bbox)
|
193 |
+
|
194 |
+
# Draw bounding box
|
195 |
+
left, top, right, bottom = int(predicted_bbox[0] * w), int(predicted_bbox[1] * h), int(predicted_bbox[2] * w), int(predicted_bbox[3] * h)
|
196 |
+
cv2.rectangle(image, (left, top), (right, bottom), color, 2)
|
197 |
+
cv2.putText(image, f'Person {identity}', (left, top - 10), cv2.FONT_HERSHEY_SIMPLEX, 0.5, color, 2)
|
198 |
+
|
199 |
+
except Exception as e:
|
200 |
+
print(f"Error during processing: {str(e)}")
|
201 |
+
return image # Return original image if there's an error
|
202 |
|
203 |
return image
|
|
|
204 |
def gradio_interface(input_image):
|
205 |
+
if input_image is None:
|
206 |
+
return None
|
207 |
+
|
208 |
+
# Convert PIL Image to numpy array if necessary
|
209 |
+
if hasattr(input_image, 'convert'):
|
210 |
+
input_image = np.array(input_image.convert('RGB'))
|
211 |
+
|
212 |
# Process the input image
|
213 |
output_image = process_image(input_image)
|
214 |
return output_image
|