| from ...smp import * |
| from .multiple_choice import extract_answer_from_item |
| import numpy as np |
| import re |
|
|
| FAIL_MSG = 'Failed to obtain answer via API.' |
|
|
| DURATIONS = [ |
| "<1min", |
| "1-2min", |
| "2-4min", |
| "4-6min", |
| "6-8min", |
| ">8min" |
| ] |
|
|
| DOMAINS = [ |
| 'Tech & Science', |
| 'Culture & Politics', |
| 'Daily Life', |
| 'Film & TV', |
| 'Performance', |
| 'Games', |
| 'Sports', |
| 'Music', |
| ] |
|
|
| SUB_CATEGORIES = [ |
| "Academic Lectures", |
| "Auto", |
| "Software", |
| "Physics", |
| "Climate Change", |
| "Space Missions", |
| "Chemistry", |
| "Engineering Projects", |
| "Biology", |
| "Science Explainers", |
| "Artificial Intelligence", |
| "Astronomy", |
| "Tech Reviews", |
| "Editorials", |
| "Politics", |
| "Historical Analysis", |
| "Social Commentary", |
| "Book Reviews", |
| "Cultural Explainers", |
| "Drawing Tutorials", |
| "Celebrity Interviews", |
| "Art Exhibitions", |
| "Fashion", |
| "Travel", |
| "Daily Vlogs", |
| "Cooking", |
| "Pranks", |
| "Camping", |
| "Nutrition & Health", |
| "Home Improvement", |
| "Painting & Photography", |
| "Unboxing Videos", |
| "Family Vlogs", |
| "DIY & Crafts", |
| "Skincare & Makeup", |
| "Documentaries", |
| "Film Trailers", |
| "Event Livestreams", |
| "Short Films", |
| "Documentary Profiles", |
| "Movie Reviews", |
| "World News", |
| "Talks", |
| "Parodies", |
| "Storytime", |
| "Stand-up", |
| "Sketches", |
| "FPS Game", |
| "Casual Game", |
| "Role Playing Game", |
| "Sports Game", |
| "Basketball", |
| "Racing", |
| "Football", |
| "Bowling Ball", |
| "Soccer", |
| "Motorsport", |
| "swimming", |
| "Boxing", |
| "Other Sports", |
| "Fitness", |
| "Fishing", |
| "Hiking", |
| "Covers", |
| "Music Videos", |
| "Remixes", |
| "Walkthroughs" |
| ] |
|
|
| TASK_DOMAINS = [ |
| 'Recognition', |
| 'Understanding', |
| 'Reasoning' |
| ] |
|
|
| TASK_CATEGORIES = [ |
| "Anomaly Recognition", |
| "Event Recognition", |
| "Attribute Recognition", |
| "Human Interaction", |
| "Temporal Localization", |
| "Video Emotions", |
| "Event Sorting", |
| "Hallucination", |
| "Text and Diagram Understanding", |
| "Attribute Reasoning", |
| "Causal Reasoning", |
| "Object Counting", |
| "Action Counting", |
| "Temporal Prediction", |
| "Emotion Change", |
| "Audio Counting", |
| "Scene Recognition", |
| "Human-object Interaction", |
| "Human Emotions", |
| "Object State Change", |
| "Relation Reasoning", |
| "Spatial Relation", |
| "Audio Source Localization", |
| "Audio Recognition", |
| "Object Existence Recognition", |
| "Audio Change" |
| ] |
|
|
| AUDIO_CLASSES = [ |
| "Speech", |
| "Event", |
| "Music", |
| ] |
|
|
|
|
| def get_dimension_rating(data_path): |
| data = load(data_path) |
|
|
| duration_rating = {k: {} for k in DURATIONS} |
| for duration in DURATIONS + ['overall']: |
| duration_rating[duration] = { |
| 'overall': '', |
| 'domain': {k: [] for k in DOMAINS}, |
| 'sub_category': {k: [] for k in SUB_CATEGORIES}, |
| 'task_domain': {k: [] for k in TASK_DOMAINS}, |
| 'task_type': {k: [] for k in TASK_CATEGORIES}, |
| 'audio_class': {k: [] for k in AUDIO_CLASSES}, |
| } |
|
|
| for i in range(len(data)): |
|
|
| domain = data.iloc[i]['domain'] |
| sub_ctg = data.iloc[i]['sub_category'] |
| task_domain_ctg = data.iloc[i]['task_domain'] |
| task_ctg = data.iloc[i]['task_type'] |
| audio_ctg = eval(data.iloc[i]['audio_class']) |
|
|
| duration = data.iloc[i]['duration'] |
| score = float(data.iloc[i]['score']) |
|
|
| duration_rating['overall']['domain'][domain].append(score) |
| duration_rating['overall']['sub_category'][sub_ctg].append(score) |
| duration_rating['overall']['task_domain'][task_domain_ctg].append(score) |
| duration_rating['overall']['task_type'][task_ctg].append(score) |
|
|
| duration_rating[duration]['domain'][domain].append(score) |
| duration_rating[duration]['sub_category'][sub_ctg].append(score) |
| duration_rating[duration]['task_domain'][task_domain_ctg].append(score) |
| duration_rating[duration]['task_type'][task_ctg].append(score) |
|
|
| for _audio_ctg in audio_ctg: |
| duration_rating['overall']['audio_class'][_audio_ctg].append(score) |
| duration_rating[duration]['audio_class'][_audio_ctg].append(score) |
| |
| for duration in ['overall'] + DURATIONS: |
|
|
| overall_res_dur = f'{np.mean([x for x in sum(duration_rating[duration]["domain"].values(), []) if x >= 0]):.3f}' |
| duration_rating[duration]['overall'] = overall_res_dur |
|
|
| for domain in DOMAINS: |
| domain_res_dur = f'{np.mean([x for x in duration_rating[duration]["domain"][domain] if x >= 0]):.3f}' |
| duration_rating[duration]['domain'][domain] = domain_res_dur |
|
|
| for sub_ctg in SUB_CATEGORIES: |
| sub_res_dur = f'{np.mean([x for x in duration_rating[duration]["sub_category"][sub_ctg] if x >= 0]):.3f}' |
| duration_rating[duration]['sub_category'][sub_ctg] = sub_res_dur |
|
|
| for task_ctg in TASK_DOMAINS: |
| task_res_dur = f'{np.mean([x for x in duration_rating[duration]["task_domain"][task_ctg] if x >= 0]):.3f}' |
| duration_rating[duration]['task_domain'][task_ctg] = task_res_dur |
| |
| for task_ctg in TASK_CATEGORIES: |
| task_res_dur = f'{np.mean([x for x in duration_rating[duration]["task_type"][task_ctg] if x >= 0]):.3f}' |
| duration_rating[duration]['task_type'][task_ctg] = task_res_dur |
|
|
| for audio_ctg in AUDIO_CLASSES: |
| audio_res_dur = f'{np.mean([x for x in duration_rating[duration]["audio_class"][audio_ctg] if x >= 0]):.3f}' |
| duration_rating[duration]['audio_class'][audio_ctg] = audio_res_dur |
|
|
| return duration_rating |
|
|
|
|
| def extract_option(model, input_item, dataset_name): |
| options = input_item['question'].split('\n')[1:] |
| for id, option in enumerate(options): |
| option_id = chr(ord('A') + id) + '.' |
| if option.find(option_id) >= 0: |
| input_item[chr(ord('A') + id)] = option[option.find(option_id) + len(option_id):].strip('. \n') |
| return extract_answer_from_item(model, input_item, dataset_name)['opt'] |
|
|
|
|
| def extract_characters_regex(s): |
| s = s.strip() |
| answer_prefixes = [ |
| 'The best answer is', |
| 'The correct answer is', |
| 'The answer is', |
| 'The answer', |
| 'The best option is' |
| 'The correct option is', |
| 'Best answer:' |
| 'Best option:', |
| 'Answer:', |
| 'Option:', |
| ] |
| for answer_prefix in answer_prefixes: |
| s = s.replace(answer_prefix, '') |
|
|
| if len(s.split()) > 10 and not re.search('[ABCD]', s): |
| return '' |
| matches = re.search(r'[ABCD]', s) |
| if matches is None: |
| return '' |
| return matches[0] |
|
|
