import gradio as gr # Gradio package for interface import sys # System package for path dependencies sys.path.append('Interface_Dependencies') sys.path.append('Engineering-Clinic-Emerging-AI-Design-Interface/Interface_Dependencies') sys.path.append('Engineering-Clinic-Emerging-AI-Design-Interface/yolov7-main') sys.path.append('./') # to run '$ python *.py' files in subdirectories from run_methods import run_all, correct_video # Gradio Interface Code with gr.Blocks(title="yolov7 Interface",theme=gr.themes.Base()) as demo: gr.Markdown( """ # Image & Video Interface for yolov7 Model Upload your own image or video and watch yolov7 try to guess what it is! """) # For for input & output settings with gr.Row() as file_settings: # Allows choice for uploading image or video [for all] file_type = gr.Radio(label="File Type",info="Choose 'Image' if you are uploading an image, Choose 'Video' if you are uploading a video", choices=['Image','Video'],value='Image',show_label=True,interactive=True,visible=True) # Allows choice of source, from computer or webcam [for all] source_type = gr.Radio(label="Source Type",info="Choose 'Computer' if you are uploading from your computer, Choose 'Webcam' if you would like to use your webcam", choices=['Computer','Webcam'],value='Computer',show_label=True,interactive=True,visible=True) # Allows choice of which convolutional layer to show (1-17) [only for images] conv_layer = gr.Slider(label="Convolution Layer",info="Choose a whole number from 1 to 17 to see the corresponding convolutional layer", minimum=1,maximum=17,value=1,interactive=True,step=1,show_label=True) # Allows choice if video from webcam is streaming or uploaded [only for webcam videos] video_stream = gr.Checkbox(label="Stream from webcam?",info="Check this box if you would like to stream from your webcam",value=False,show_label=True,interactive=True,visible=False) # Allows choice of which smooth gradient output to show (1-3) [only for images] output_map = gr.Slider(label="Map Output Number",info="Choose a whole number from 1 to 3 to see the corresponding attribution map", minimum=1,maximum=3,value=1,interactive=True,step=1,show_label=True) # For all inputs & outputs with gr.Row() as inputs_outputs: # Default input image: Visible, Upload from computer input_im = gr.Image(source="upload",type='filepath',label="Input Image", show_download_button=True,show_share_button=True,interactive=True,visible=True) # Default Boxed output image: Visible output_box_im = gr.Image(type='filepath',label="Output Image", show_download_button=True,show_share_button=True,interactive=False,visible=True) # Defualt Convolutional output image: Visible output_conv_im = gr.Image(type='filepath',label="Output Convolution", show_download_button=True,show_share_button=True,interactive=False,visible=True) # Default Gradient output image: Visible output_grad_im = gr.Image(type='filepath',label="Output Smooth Gradient", show_download_button=True,show_share_button=True,interactive=False,visible=True) # Default label output textbox: Visible labels = gr.Textbox(label='Top Predictions', value = "") # Default time output textbox: Visible formatted_time = gr.Textbox(label = 'Time to Run in Seconds:', value = "") # Default input video: Not visible, Upload from computer input_vid = gr.Video(source="upload",label="Input Video", show_share_button=True,interactive=True,visible=False) # Default Boxed output video: Not visible output_box_vid = gr.Video(label="Output Video",show_share_button=True,visible=False) # List of components for clearing clear_comp_list = [input_im, output_box_im, output_conv_im, output_grad_im, labels, formatted_time, input_vid, output_box_vid] # For start & clear buttons with gr.Row() as buttons: start_but = gr.Button(label="Start") clear_but = gr.ClearButton(value='Clear All',components=clear_comp_list, interactive=True,visible=True) # For model settings with gr.Row() as model_settings: # Pixel size of the inference [Possibly useless, may remove] inf_size = gr.Number(label='Inference Size (pixels)',value=640,precision=0) # Object confidence threshold obj_conf_thr = gr.Number(label='Object Confidence Threshold',value=0.25) # Intersection of union threshold iou_thr = gr.Number(label='IOU threshold for NMS',value=0.45) # Agnostic NMS boolean agnostic_nms = gr.Checkbox(label='Agnostic NMS',value=True) # Normailze gradient boolean norm = gr.Checkbox(label='Normalize Gradient',value=False,visible=True) def change_file_type(file, source, is_stream): """ Changes the visible components of the gradio interface Args: file (str): Type of the file (image or video) source (str): If the file is uploaded or from webcam is_stream (bool): If the video is streaming or uploaded Returns: Dictionary: Each component of the interface that needs to be updated. """ if file == "Image": if source == "Computer": return { conv_layer: gr.Slider(visible=True), video_stream: gr.Checkbox(visible=False, value=False), output_map: gr.Slider(visible=True), input_im: gr.Image(source="upload",type='filepath',label="Input Image", show_download_button=True,show_share_button=True,interactive=True,visible=True,streaming=False), output_box_im: gr.Image(visible=True), output_conv_im: gr.Image(visible=True), output_grad_im: gr.Image(visible=True), input_vid: gr.Video(visible=False), output_box_vid: gr.Video(visible=False), norm: gr.Checkbox(visible=True), labels: gr.Textbox(visible=True), formatted_time: gr.Textbox(visible=True) } elif source == "Webcam": return { conv_layer: gr.Slider(visible=True), video_stream: gr.Checkbox(visible=False, value=False), output_map: gr.Slider(visible=True), input_im: gr.Image(type='pil',source="webcam",label="Input Image", visible=True,interactive=True,streaming=False), output_box_im: gr.Image(visible=True), output_conv_im: gr.Image(visible=True), output_grad_im: gr.Image(visible=True), input_vid: gr.Video(visible=False), output_box_vid: gr.Video(visible=False), norm: gr.Checkbox(visible=True), labels: gr.Textbox(visible=True), formatted_time: gr.Textbox(visible=True) } elif file == "Video": if source == "Computer": return { conv_layer: gr.Slider(visible=False), video_stream: gr.Checkbox(visible=False, value=False), output_map: gr.Slider(visible=False), input_im: gr.Image(visible=False,streaming=False), output_box_im: gr.Image(visible=False), output_conv_im: gr.Image(visible=False), output_grad_im: gr.Image(visible=False), input_vid: gr.Video(source="upload",label="Input Video", show_share_button=True,interactive=True,visible=True), output_box_vid: gr.Video(label="Output Video",show_share_button=True,visible=True), norm: gr.Checkbox(visible=False), labels: gr.Textbox(visible=False), formatted_time: gr.Textbox(visible=False) } elif source == "Webcam": if is_stream: return { conv_layer: gr.Slider(visible=False), video_stream: gr.Checkbox(visible=True), output_map: gr.Slider(visible=False), input_im: gr.Image(type='pil',source="webcam",label="Input Image", streaming=True,visible=True,interactive=True), output_box_im: gr.Image(visible=True), output_conv_im: gr.Image(visible=False), output_grad_im: gr.Image(visible=False), input_vid: gr.Video(visible=False), output_box_vid: gr.Video(visible=False), norm: gr.Checkbox(visible=False), labels: gr.Textbox(visible=False), formatted_time: gr.Textbox(visible=False) } elif not is_stream: return { conv_layer: gr.Slider(visible=False), video_stream: gr.Checkbox(visible=True, value=False), output_map: gr.Slider(visible=False), input_im: gr.Image(visible=False,streaming=False), output_box_im: gr.Image(visible=False), output_conv_im: gr.Image(visible=False), output_grad_im: gr.Image(visible=False), input_vid: gr.Video(label="Input Video",source="webcam", show_share_button=True,interactive=True,visible=True), output_box_vid: gr.Video(label="Output Video",show_share_button=True,visible=True), norm: gr.Checkbox(visible=False), labels: gr.Textbox(visible=False), formatted_time: gr.Textbox(visible=False) } def change_conv_layer(layer): """ Changes the shown convolutional output layer based on gradio slider Args: layer (int): The layer to show Returns: str: The file path of the output image """ return "outputs\\runs\\detect\\exp\\layers\\layer" + str(int(int(layer) - 1)) + '.jpg' def change_output_num(number): return "outputs\\runs\\detect\\exp\\smoothGrad" + str(int(int(number) -1)) + '.jpg' # List of gradio components that change during method "change_file_type" change_comp_list = [conv_layer, video_stream, output_map, input_im, output_box_im, output_conv_im, output_grad_im, input_vid, output_box_vid, norm, labels, formatted_time] # List of gradio components that are input into the run_all method (when start button is clicked) run_inputs = [file_type, input_im, input_vid, source_type, inf_size, obj_conf_thr, iou_thr, conv_layer, agnostic_nms, output_map, video_stream, norm] # List of gradio components that are output from the run_all method (when start button is clicked) run_outputs = [output_box_im, output_conv_im, output_grad_im, labels, formatted_time, output_box_vid] # When these settings are changed, the change_file_type method is called file_type.input(change_file_type, show_progress=True, inputs=[file_type, source_type, video_stream], outputs=change_comp_list) source_type.input(change_file_type, show_progress=True, inputs=[file_type, source_type, video_stream], outputs=change_comp_list) video_stream.input(change_file_type, show_progress=True, inputs=[file_type, source_type, video_stream], outputs=change_comp_list) # When start button is clicked, the run_all method is called start_but.click(run_all, inputs=run_inputs, outputs=run_outputs) # When video is uploaded, the correct_video method is called input_vid.upload(correct_video, inputs=[input_vid], outputs=[input_vid]) # When the convolutional layer setting is changed, the change_conv_layer method is called conv_layer.input(change_conv_layer, conv_layer, output_conv_im) # When the stream setting is true, run the stream input_im.stream(run_all, inputs=run_inputs, outputs=run_outputs) # When the gradient number is changed, the change_output_num method is called output_map.input(change_output_num, output_map, output_grad_im) # When the demo is first started, run the change_file_type method to ensure default settings demo.load(change_file_type, show_progress=True, inputs=[file_type, source_type, video_stream], outputs=change_comp_list) if __name__== "__main__" : # If True, it launches Gradio interface # If False, it runs without the interface if True: # demo.queue().launch(share=True) demo.queue().launch() else: # run_image("inference\\images\\bus.jpg","Computer",640,0.45,0.25,1,True) run_video("0", "Webcam", 640, 0.25, 0.45, True, True)