Spaces:
Build error
Build error
| import gradio as gr | |
| import numpy as np | |
| import cv2 | |
| # model load | |
| cfg = r'volleyball_test.cfg' | |
| weights = r'volleyball_final.weights' | |
| net = cv2.dnn.readNetFromDarknet(cfg, weights) | |
| # classes | |
| classes = [] | |
| with open("classes.names", 'r') as f: | |
| classes = f.read().splitlines() | |
| def predict_img(img_bgr): | |
| # img_bgr = inp.astype('uint8')[...,::-1] | |
| img = cv2.resize(img_bgr, (700, 700)) | |
| height, width, channels = img.shape | |
| # Convert image into blob and load it on model | |
| blob = cv2.dnn.blobFromImage( | |
| img, 1/255, (height, width), (0, 0, 0), swapRB=True, crop=False) | |
| net.setInput(blob) | |
| # Getting all the three detection layers of yolo | |
| output_layers_names = net.getUnconnectedOutLayersNames() | |
| # print(output_layers_names) | |
| layersOutputs = net.forward(output_layers_names) | |
| # print(layersOutputs) | |
| # Finding the y-vector and minimum no.of bounding box | |
| confthreshold = 0.5 | |
| boxes = [] | |
| confidences = [] | |
| class_ids = [] | |
| for output in layersOutputs: | |
| for detection in output: | |
| scores = detection[5:] | |
| class_id = np.argmax(scores) | |
| confidence = scores[class_id] | |
| if confidence > confthreshold: | |
| center_x = int(detection[0]*width) | |
| center_y = int(detection[1]*height) | |
| w = int(detection[2]*width) | |
| h = int(detection[3]*height) | |
| x = int(center_x - w/2) | |
| y = int(center_y - h/2) | |
| boxes.append([x, y, w, h]) | |
| confidences.append(float(confidence)) | |
| class_ids.append(class_id) | |
| # Applying Non max Suppression for removing unwanted multiple bounding boxes | |
| indexes = cv2.dnn.NMSBoxes( | |
| boxes, confidences, confthreshold, nms_threshold=0.3) | |
| for i in indexes: | |
| box = boxes[i] | |
| x, y, w, h = box | |
| cv2.rectangle(img, (x, y), (x+w, y+h), (0, 255, 0), 2) | |
| conf_value = str(round(confidences[i], 2)) | |
| label = str(classes[class_ids[i]]) | |
| cv2.putText(img, label + " " + conf_value, (x, y-10), | |
| cv2.FONT_HERSHEY_COMPLEX, 1, (0, 255, 0), 2) | |
| # return the images | |
| # cv2.imwrite("out.jpg", img) | |
| return img | |
| def predict(inp): | |
| vidcap = cv2.VideoCapture(inp) | |
| fps = vidcap.get(cv2.CAP_PROP_FPS) | |
| outcap = cv2.VideoWriter('outpy.mp4',cv2.VideoWriter_fourcc(*'MP4V'), fps, (700, 700)) | |
| success,image = vidcap.read() | |
| count = 0 | |
| while success: | |
| img = predict_img(image) | |
| outcap.write(img) | |
| success,image = vidcap.read() | |
| print('Read a new frame: ', success) | |
| count += 1 | |
| return "./outpy.mp4" | |
| gr.Interface( | |
| fn=predict, | |
| inputs=[ | |
| gr.inputs.Video() # you can have many inputs | |
| ], | |
| outputs=[ | |
| gr.inputs.Video() # you can have many outputs | |
| ], | |
| title="Volley classification and detection", | |
| description="This project use a yolov3 and pretrained model from [this](https://github.com/lalchhabi/Volleyball_Position_Detection_System) project", | |
| examples=[ | |
| "test.mp4", | |
| ] | |
| ).launch() | |