import gradio as gr import numpy as np import json from zipfile import ZipFile import os from backend.dataloader import create_dataloader_aris from backend.aws_handler import ping_server from backend.predict import predict_task from backend.uploader import save_data_to_dir, create_data_dir, save_data from backend.InferenceConfig import InferenceConfig from frontend.upload_ui import Upload_Gradio, models from frontend.result_ui import Result_Gradio, update_result, create_metadata_table, table_headers, info_headers from frontend.annotation_handler import load_annotation, prepare_annotation, js_store_frame_info, annotation_css from frontend.state_handler import reset_state #Initialize State & Result state = { 'version': "Experimental 1.0", 'files': [], 'index': 1, 'total': 1, 'annotation_index': -1, 'frame_index': 0, 'outputs': [], 'config': None, 'enable_annotation_editor': False, } result = {} components = {} # -------------------------------------------- ----- UPLOAD ARIS FILE ------------------------------------------------------ # Called when an Aris file is uploaded for inference - calls infer_next def on_aris_input( file_list, model_id, conf_thresh, iou_thresh, min_hits, max_age, associative_tracker, boost_power, boost_decay, byte_low_conf, byte_high_conf, min_length, max_length, min_travel, output_formats ): if isinstance(file_list, tuple): file_list = [file_list] print(output_formats) # Reset Result reset_state(result, state) state['files'] = file_list state['total'] = len(file_list) state['outputs'] = output_formats state['config'] = InferenceConfig( weights = models[model_id] if model_id in models else models['master'], conf_thresh = conf_thresh, nms_iou = iou_thresh, min_hits = min_hits, max_age = max_age, min_length = min_length, max_length = max_length, min_travel = min_travel, ) # Enable tracker if specified if (associative_tracker == "Confidence Boost"): state['config'].enable_conf_boost(boost_power, boost_decay) elif (associative_tracker == "ByteTrack"): state['config'].enable_byte_track(byte_low_conf, byte_high_conf) else: state['config'].enable_sort_track() print(" ") print("Inference with:") print(state['config'].to_dict()) print(" ") # Update loading_space to start inference on first file return { inference_handler: gr.update(value = str(np.random.rand()), visible=True), components['cancel_btn']: gr.update(visible=True), master_tabs: gr.update(selected=1) } # Iterative function that performs inference on the next file in line def infer_next(_, progress=gr.Progress()): if state['index'] >= state['total']: return { result_handler: gr.update(), inference_handler: gr.update() } # Correct progress function for batch file input set_progress = lambda pct, msg : progress(pct, desc=msg) if state['total'] > 1: set_progress = lambda pct, msg : progress(pct, desc="File " + str(state['index']+1) + "/" + str(state['total']) + ": " + msg) set_progress(0, "Starting...") # Save file and create a new directory for result file_info = state['files'][state['index']] file_name = file_info[0].split("/")[-1] bytes = file_info[1] valid, file_path, dir_name = save_data(bytes, file_name) print("Directory: ", dir_name) print("Aris input: ", file_path) print(" ") # Check that the file was valid if not valid: return { result_handler: gr.update(), inference_handler: gr.update() } # Send uploaded file to AWS ping_server(file_name, state) #upload_file(file_path, "fishcounting", "webapp_uploads/files/" + file_name) #crop_clip(file_path, 65) # Do inference json_result, json_filepath, zip_filepath, video_filepath, marking_filepath = predict_task( file_path, config = state['config'], output_formats = state['outputs'], gradio_progress = set_progress ) # prepare dummy dataloader for visualizations _, dataset = create_dataloader_aris(file_path, num_frames_bg_subtract=0) # Store result for that file result['json_result'].append(json_result) result['aris_input'].append(file_path) result['datasets'].append(dataset) result["path_video"].append(video_filepath) result["path_zip"].append(zip_filepath) result["path_json"].append(json_filepath) result["path_marking"].append(marking_filepath) fish_table, fish_info = create_metadata_table(json_result, table_headers, info_headers) result["fish_table"].append(fish_table) result["fish_info"].append(fish_info) # Increase file index state['index'] += 1 # Send of update to result_handler to show new result # Leave inference_handler update blank to avoid starting next inference until result is updated return { result_handler: gr.update(value = str(np.random.rand())), tab_labeler: gr.update(value = str(state['index'])), inference_handler: gr.update() } # Cancel inference def cancel_inference(): return { master_tabs: gr.update(selected=0), inference_handler: gr.update(visible=False), components['cancel_btn']: gr.update(visible=False) } # Show result def on_result_ready(): # Update result tab for last file i = state["index"] - 1 return update_result(i, state, result, inference_handler) # ------------------------------------------------- UPLOAD RESULT FILE ----------------------------------------------------- # Called when result file is uploaded for review def on_result_upload(): return { master_tabs: gr.update(selected=1), result_uploader: gr.update(value=str(np.random.rand())) } # Called when result upload is finished processing def on_result_upload_finish(zip_list, aris_list): if (zip_list == None): zip_list = [("static/example/example_result.zip", None)] aris_path = "static/example/input_file.aris" aris_list = [(aris_path, bytearray(open(aris_path, 'rb').read()))] reset_state(result, state) state['outputs'] = ["Generate Annotated Video", "Generate Manual Marking", "Generate PDF"] component_updates = { tab_labeler: gr.update(value = len(zip_list)) } for i in range(len(zip_list)): # Create dir to unzip files dir_name = create_data_dir(str(i)) # Check aris input if (aris_list): aris_info = aris_list[i] file_name = aris_info[0].split("/")[-1] bytes = aris_info[1] valid, input_path, dir_name = save_data_to_dir(bytes, file_name, dir_name) _, dataset = create_dataloader_aris(input_path, num_frames_bg_subtract=0) else: input_path = None dataset = None # Unzip result zip_info = zip_list[i] zip_name = zip_info[0] print(zip_name) with ZipFile(zip_name) as zip_file: ZipFile.extractall(zip_file, path=dir_name) unzipped = os.listdir(dir_name) print(unzipped) for file in unzipped: if (file.endswith("_results.mp4")): result["path_video"].append(os.path.join(dir_name, file)) elif (file.endswith("_results.json")): result["path_json"].append(os.path.join(dir_name, file)) elif (file.endswith("_marking.txt")): result["path_marking"].append(os.path.join(dir_name, file)) result["aris_input"].append(input_path) result["datasets"].append(dataset) with open(result['path_json'][-1]) as f: json_result = json.load(f) result['json_result'].append(json_result) fish_table, fish_info = create_metadata_table(json_result, table_headers, info_headers) result["fish_table"].append(fish_table) result["fish_info"].append(fish_info) update = update_result(i, state, result, inference_handler) for key in update.keys(): component_updates[key] = update[key] component_updates.pop(inference_handler) return component_updates # ------------------------------------------------- ANNOTATION EDITOR ----------------------------------------------------- def on_annotation_open(result_index): return prepare_annotation(state, result, result_index) def annotate_next(_, progress=gr.Progress()): return load_annotation(state, result, progress) # -------------------------------------------------- GRADIO ARCHITECTURE ---------------------------------------------------- with gr.Blocks() as demo: with gr.Blocks(): # Title of page + style gr.HTML( """

Caltech Fisheye - Experimental

""" ) with gr.Tabs() as master_tabs: components['master_tabs'] = master_tabs # Master Tab for uploading aris or result files with gr.Tab("Upload", id=0): # Draw Gradio components related to the upload ui Upload_Gradio(components) # Master Tab for result visualization with gr.Tab("Result", id=1): # Define annotation progress bar for event listeres, but unrender since it will be displayed later on result_uploader = gr.HTML("", visible=False) components['result_uploader'] = result_uploader annotation_progress = gr.HTML("", visible=False).unrender() components['annotation_progress'] = annotation_progress # Draw the gradio components related to visualzing result visualization_components = Result_Gradio(on_annotation_open, components, state) # Master Tab for annotation editing if state['enable_annotation_editor']: with gr.Tab("Annotation Editor", id=2): # Draw the annotation loading bar here annotation_progress.render() # Add annotation editor component annotation_editor = gr.HTML("", visible=False) # Event listener for batch loading of annotation frames annotation_progress.change( annotate_next, annotation_progress, [annotation_editor, annotation_progress], _js=js_store_frame_info ) # Event listener for running javascript defined in 'annotation_editor.js' # show_annotation with open('gradio_scripts/annotation_editor.js', 'r') as f: annotation_editor.change(lambda x: gr.update(), None, annotation_editor, _js=f.read()) # Disclaimer at the bottom of page gr.HTML( """

Note: The software is provided "as is", without warranty of any kind, express or implied, including but not limited to the warranties of merchantability, fitness for a particular purpose and noninfringement. In no event shall the authors or copyright holders be liable for any claim, damages or other liability, whether in an action of contract, tort or otherwise, arising from, out of or in connection with the software or the use or other dealings in the software.

""" ) # Extract important components for ease of code input = components['input'] inference_handler = components['inference_handler'] result_handler = components['result_handler'] tab_labeler = components['tab_labeler'] inference_comps = [inference_handler, master_tabs, components['cancel_btn']] # When a file is uploaded to the input, tell the inference_handler to start inference input.upload(on_aris_input, [input] + components['hyperparams'], inference_comps) components['inference_btn'].click(on_aris_input, [input] + components['hyperparams'], inference_comps) # When inference handler updates, tell result_handler to show the new result # Also, add inference_handler as the output in order to have it display the progress inference_event = inference_handler.change(infer_next, None, [inference_handler, result_handler, tab_labeler]) # Send UI changes based on the new results to the UI_components, and tell the inference_handler to start next inference result_handler.change(on_result_ready, None, visualization_components + [inference_handler]) # Cancel and skip buttons components['cancel_btn'].click(cancel_inference, None, inference_comps, cancels=[inference_event]) # Button to load a previous result and view visualization components['open_result_btn'].click(on_result_upload, None, [result_uploader, master_tabs]) components['result_uploader'].change( on_result_upload_finish, [components['result_input'], components['result_aris_input']], visualization_components + [tab_labeler] ) demo.queue().launch()