{"cells":[{"metadata":{"trusted":true},"cell_type":"code","source":"!pip uninstall torch --y\n!pip uninstall torchaudio --y","execution_count":null,"outputs":[]},{"metadata":{"_cell_guid":"b1076dfc-b9ad-4769-8c92-a6c4dae69d19","_uuid":"8f2839f25d086af736a60e9eeb907d3b93b6e0e5","execution":{"iopub.execute_input":"2020-08-11T09:45:24.805587Z","iopub.status.busy":"2020-08-11T09:45:24.80465Z","iopub.status.idle":"2020-08-11T09:45:28.87251Z","shell.execute_reply":"2020-08-11T09:45:28.871377Z"},"papermill":{"duration":4.091991,"end_time":"2020-08-11T09:45:28.872641","exception":false,"start_time":"2020-08-11T09:45:24.78065","status":"completed"},"tags":[],"trusted":true},"cell_type":"code","source":"# Newest torch required for tracing. Torchaudio for data loading\n!pip install ../input/pytorch-160-with-torchvision-070/torch-1.6.0cu101-cp37-cp37m-linux_x86_64.whl\n!pip install ../input/torchaudio/torchaudio-0.6.0-cp37-cp37m-manylinux1_x86_64.whl\n# !pip install ../input/bird-panns/torchlibrosa-master/torchlibrosa-master/ > /dev/null\n\nimport os\nimport gc\nimport time\nimport math\nimport shutil\nimport random\nimport warnings\nimport typing as tp\nfrom pathlib import Path\nfrom contextlib import contextmanager\n\nimport yaml\nfrom joblib import delayed, Parallel\n\nimport cv2\nimport librosa\nimport audioread\nimport soundfile as sf\n\nimport numpy as np\nimport pandas as pd\n\nfrom fastprogress import progress_bar\nfrom sklearn.metrics import f1_score\nfrom sklearn.model_selection import StratifiedKFold\n\nimport torch\nimport torchaudio\nimport torch.nn as nn\nimport torch.nn.functional as F\nfrom torch.nn import Conv2d, Module, Linear, BatchNorm2d, ReLU\nfrom torch.nn.modules.utils import _pair\nimport torch.utils.data as data\n#from torchlibrosa.stft import Spectrogram, LogmelFilterBank\n#from torchlibrosa.augmentation import SpecAugmentation\n#from efficientnet_pytorch import EfficientNet\n\n\npd.options.display.max_rows = 500\npd.options.display.max_columns = 500","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"import os","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"sorted(os.listdir('../input/birdweights'))","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"args = {\n    'n_mels': 224,\n    'hop_length': 384,\n    'std_sr': 32000,\n    'threshold': .5,\n    'model': '../input/birdweights/c2__17.pt',  # c2__17.pt c2_9_18.pt, c2_0_41.pt\n    'model2': '../input/birdweights/c2_9_18.pt',\n    'model3': '../input/birdweights/c2_0_41.pt',\n    #'model5': '../input/birdweights/c3_4_17.pt',\n    #'model6': '../input/birdweights/c6_1_54.pt',\n    #'model7': '../input/birdweights/c7_0_30.pt',\n    #'model8': '../input/birdweights/leapoffaith.pt'\n}\nmodel = torch.jit.load(args['model'], map_location='cpu').cuda()\nmodel.eval();\nmodel2 = torch.jit.load(args['model2'], map_location='cpu').cuda()\nmodel2.eval();\nmodel3 = torch.jit.load(args['model3'], map_location='cpu').cuda()\nmodel3.eval();\n# model5 = torch.jit.load(args['model5'], map_location='cpu').cuda()\n# model5.eval();\n# model6 = torch.jit.load(args['model6'], map_location='cpu').cuda()\n# model6.eval();\n# model7 = torch.jit.load(args['model7'], map_location='cpu').cuda()\n# model7.eval();\n# model8 = torch.jit.load(args['model8'], map_location='cpu').cuda()\n# model8.eval();\n\nroot = os.path.join('../input', \"birdsong-recognition\")\n# TRAIN_AUDIO_DIR = RAW_DATA / \"train_audio\"\ntest_dir = os.path.join(root, 'test_audio')\nprint(test_dir)\n\nif os.path.isdir(test_dir):\n    print('Running actual submission!')\n    test_df = pd.read_csv(os.path.join(root, 'test.csv'))\n    sub = pd.read_csv(\"../input/birdsong-recognition/sample_submission.csv\")\nelse:\n    print('Running debug submission')\n    test_dir = os.path.join('../input', 'birdcall-check', 'test_audio')\n    test_df = pd.read_csv(os.path.join('../input', 'birdcall-check', 'test.csv'))\n    sub = pd.read_csv('../input/inference-pytorch-birdcall-resnet-baseline/submission.csv')\n    print('Naturally no test dir found')\nsub.to_csv(\"submission.csv\", index=False)  # this will be overwritten if everything goes well. Hmnmm. will it?","execution_count":null,"outputs":[]},{"metadata":{"_kg_hide-input":true,"execution":{"iopub.execute_input":"2020-08-11T09:45:29.471784Z","iopub.status.busy":"2020-08-11T09:45:29.470935Z","iopub.status.idle":"2020-08-11T09:45:29.474035Z","shell.execute_reply":"2020-08-11T09:45:29.474607Z"},"papermill":{"duration":0.04791,"end_time":"2020-08-11T09:45:29.474766","exception":false,"start_time":"2020-08-11T09:45:29.426856","status":"completed"},"tags":[],"trusted":true},"cell_type":"code","source":"BIRD_CODE = {\n    'aldfly': 0, 'ameavo': 1, 'amebit': 2, 'amecro': 3, 'amegfi': 4,\n    'amekes': 5, 'amepip': 6, 'amered': 7, 'amerob': 8, 'amewig': 9,\n    'amewoo': 10, 'amtspa': 11, 'annhum': 12, 'astfly': 13, 'baisan': 14,\n    'baleag': 15, 'balori': 16, 'banswa': 17, 'barswa': 18, 'bawwar': 19,\n    'belkin1': 20, 'belspa2': 21, 'bewwre': 22, 'bkbcuc': 23, 'bkbmag1': 24,\n    'bkbwar': 25, 'bkcchi': 26, 'bkchum': 27, 'bkhgro': 28, 'bkpwar': 29,\n    'bktspa': 30, 'blkpho': 31, 'blugrb1': 32, 'blujay': 33, 'bnhcow': 34,\n    'boboli': 35, 'bongul': 36, 'brdowl': 37, 'brebla': 38, 'brespa': 39,\n    'brncre': 40, 'brnthr': 41, 'brthum': 42, 'brwhaw': 43, 'btbwar': 44,\n    'btnwar': 45, 'btywar': 46, 'buffle': 47, 'buggna': 48, 'buhvir': 49,\n    'bulori': 50, 'bushti': 51, 'buwtea': 52, 'buwwar': 53, 'cacwre': 54,\n    'calgul': 55, 'calqua': 56, 'camwar': 57, 'cangoo': 58, 'canwar': 59,\n    'canwre': 60, 'carwre': 61, 'casfin': 62, 'caster1': 63, 'casvir': 64,\n    'cedwax': 65, 'chispa': 66, 'chiswi': 67, 'chswar': 68, 'chukar': 69,\n    'clanut': 70, 'cliswa': 71, 'comgol': 72, 'comgra': 73, 'comloo': 74,\n    'commer': 75, 'comnig': 76, 'comrav': 77, 'comred': 78, 'comter': 79,\n    'comyel': 80, 'coohaw': 81, 'coshum': 82, 'cowscj1': 83, 'daejun': 84,\n    'doccor': 85, 'dowwoo': 86, 'dusfly': 87, 'eargre': 88, 'easblu': 89,\n    'easkin': 90, 'easmea': 91, 'easpho': 92, 'eastow': 93, 'eawpew': 94,\n    'eucdov': 95, 'eursta': 96, 'evegro': 97, 'fiespa': 98, 'fiscro': 99,\n    'foxspa': 100, 'gadwal': 101, 'gcrfin': 102, 'gnttow': 103, 'gnwtea': 104,\n    'gockin': 105, 'gocspa': 106, 'goleag': 107, 'grbher3': 108, 'grcfly': 109,\n    'greegr': 110, 'greroa': 111, 'greyel': 112, 'grhowl': 113, 'grnher': 114,\n    'grtgra': 115, 'grycat': 116, 'gryfly': 117, 'haiwoo': 118, 'hamfly': 119,\n    'hergul': 120, 'herthr': 121, 'hoomer': 122, 'hoowar': 123, 'horgre': 124,\n    'horlar': 125, 'houfin': 126, 'houspa': 127, 'houwre': 128, 'indbun': 129,\n    'juntit1': 130, 'killde': 131, 'labwoo': 132, 'larspa': 133, 'lazbun': 134,\n    'leabit': 135, 'leafly': 136, 'leasan': 137, 'lecthr': 138, 'lesgol': 139,\n    'lesnig': 140, 'lesyel': 141, 'lewwoo': 142, 'linspa': 143, 'lobcur': 144,\n    'lobdow': 145, 'logshr': 146, 'lotduc': 147, 'louwat': 148, 'macwar': 149,\n    'magwar': 150, 'mallar3': 151, 'marwre': 152, 'merlin': 153, 'moublu': 154,\n    'mouchi': 155, 'moudov': 156, 'norcar': 157, 'norfli': 158, 'norhar2': 159,\n    'normoc': 160, 'norpar': 161, 'norpin': 162, 'norsho': 163, 'norwat': 164,\n    'nrwswa': 165, 'nutwoo': 166, 'olsfly': 167, 'orcwar': 168, 'osprey': 169,\n    'ovenbi1': 170, 'palwar': 171, 'pasfly': 172, 'pecsan': 173, 'perfal': 174,\n    'phaino': 175, 'pibgre': 176, 'pilwoo': 177, 'pingro': 178, 'pinjay': 179,\n    'pinsis': 180, 'pinwar': 181, 'plsvir': 182, 'prawar': 183, 'purfin': 184,\n    'pygnut': 185, 'rebmer': 186, 'rebnut': 187, 'rebsap': 188, 'rebwoo': 189,\n    'redcro': 190, 'redhea': 191, 'reevir1': 192, 'renpha': 193, 'reshaw': 194,\n    'rethaw': 195, 'rewbla': 196, 'ribgul': 197, 'rinduc': 198, 'robgro': 199,\n    'rocpig': 200, 'rocwre': 201, 'rthhum': 202, 'ruckin': 203, 'rudduc': 204,\n    'rufgro': 205, 'rufhum': 206, 'rusbla': 207, 'sagspa1': 208, 'sagthr': 209,\n    'savspa': 210, 'saypho': 211, 'scatan': 212, 'scoori': 213, 'semplo': 214,\n    'semsan': 215, 'sheowl': 216, 'shshaw': 217, 'snobun': 218, 'snogoo': 219,\n    'solsan': 220, 'sonspa': 221, 'sora': 222, 'sposan': 223, 'spotow': 224,\n    'stejay': 225, 'swahaw': 226, 'swaspa': 227, 'swathr': 228, 'treswa': 229,\n    'truswa': 230, 'tuftit': 231, 'tunswa': 232, 'veery': 233, 'vesspa': 234,\n    'vigswa': 235, 'warvir': 236, 'wesblu': 237, 'wesgre': 238, 'weskin': 239,\n    'wesmea': 240, 'wessan': 241, 'westan': 242, 'wewpew': 243, 'whbnut': 244,\n    'whcspa': 245, 'whfibi': 246, 'whtspa': 247, 'whtswi': 248, 'wilfly': 249,\n    'wilsni1': 250, 'wiltur': 251, 'winwre3': 252, 'wlswar': 253, 'wooduc': 254,\n    'wooscj2': 255, 'woothr': 256, 'y00475': 257, 'yebfly': 258, 'yebsap': 259,\n    'yehbla': 260, 'yelwar': 261, 'yerwar': 262, 'yetvir': 263\n}\n\nINV_BIRD_CODE = {v: k for k, v in BIRD_CODE.items()}","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"print('keepgoing')","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"from torchaudio.transforms import *\nfmin, fmax = 25, 11000\nf1 = MelSpectrogram(sample_rate=args['std_sr'], n_fft=1024, hop_length=args['hop_length'], f_min=fmin, f_max=fmax, n_mels=args['n_mels']).cuda()\nf2 = MelSpectrogram(sample_rate=args['std_sr'], n_fft=2048, hop_length=args['hop_length'], f_min=fmin, f_max=fmax, n_mels=args['n_mels']).cuda()\nf3 = MelSpectrogram(sample_rate=args['std_sr'], n_fft=4096, hop_length=args['hop_length'], f_min=fmin, f_max=fmax, n_mels=args['n_mels']).cuda()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"class TestDataset(data.Dataset):\n    def __init__(self, df, spectrogram):\n        self.df = df\n        self.spectrogram = spectrogram\n        \n    def __len__(self):\n        return len(self.df)\n    \n    def normalize(self, spect):\n        spect = spect.contiguous()\n        miu = spect.view(spect.size(0), -1).mean(-1).unsqueeze(-1).unsqueeze(-1)\n        std = spect.view(spect.size(0), -1).std(-1).unsqueeze(-1).unsqueeze(-1) + 1e-8\n        return (spect - miu) / std / 2\n    \n    def __getitem__(self, idx):\n        sample = self.df.loc[idx, :]\n        site = sample.site\n        row_id = sample.row_id\n        \n        if site == \"site_3\":\n            y = self.spectrogram\n            len_y = y.shape[-1]\n            start = 0\n            end = args['std_sr'] * 5 // args['hop_length']\n            y_all = []\n            while len_y > start:\n                y_batch = self.normalize(y[:, :, start:end])\n                if end >= len_y:\n                    clip_length = args['std_sr'] * 5 // args['hop_length']\n                    y_batch = self.normalize(y[:, :, -clip_length:])\n                    y_all.append(y_batch)\n                    break\n                start = end\n                end = end + args['std_sr'] * 5 // args['hop_length']\n                y_all.append(y_batch)\n            y_all = torch.stack(y_all)\n            return y_all, row_id, site\n        else:\n            end_seconds = int(sample.seconds)\n            start_seconds = int(end_seconds - 5)\n            \n            start_index = args['std_sr'] * start_seconds // args['hop_length']\n            end_index = args['std_sr'] * end_seconds // args['hop_length']\n            \n            # Hmmm. so assuming the final index will always be less than length of whole sound file\n            y = self.spectrogram[:, :, start_index:end_index]\n            y = self.normalize(y)\n        return y, row_id, site\n    \n\ndef prediction_for_clip(dataset, threshold=0.5):\n    loader = data.DataLoader(dataset, batch_size=1, shuffle=False)\n    device = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\")\n    \n    model.eval()\n    model2.eval()\n    model3.eval()\n#     model5.eval()\n#     model6.eval()\n#     model7.eval()\n#     model8.eval()\n    prediction_dict = {}\n    for image, row_id, site in loader:\n        site = site[0]\n        row_id = row_id[0]\n        print('imageshape',image.shape)\n        image = image.to(device).float()\n        if site in {\"site_1\", \"site_2\"}:\n            # print(row_id, image.shape)\n            image = image.to(device).float()\n            with torch.no_grad():\n                prediction = model(image)\n                print('prediction of model',prediction.shape)\n                assert 1==2\n                proba = torch.square(prediction.view(-1).sigmoid()).cpu().numpy()\n                prediction = model2(image)\n                proba += torch.square(prediction.view(-1).sigmoid()).cpu().numpy()\n                prediction = model3(image)\n                proba += torch.square(prediction.view(-1).sigmoid()).cpu().numpy()\n#                 prediction = model5(image)\n#                 proba += torch.square(prediction.view(-1).sigmoid()).cpu().numpy()\n#                 prediction = model6(image)\n#                 proba += torch.square(prediction.view(-1).sigmoid()).cpu().numpy()\n#                 prediction = model7(image)\n#                 proba += torch.square(prediction.view(-1).sigmoid()).cpu().numpy()\n#                 prediction = model8(image)\n#                 proba += torch.square(prediction.view(-1).sigmoid()).cpu().numpy()\n                \n                proba = proba / 3\n                proba = np.sqrt(proba)\n\n            events = proba >= threshold\n            labels = np.argwhere(events).reshape(-1).tolist()\n        else:\n            # print(row_id, image.shape)\n            # to avoid prediction on large batch\n            image = image.squeeze(0)\n            batch_size = 16\n            whole_size = image.size(0)\n            if whole_size % batch_size == 0:\n                n_iter = whole_size // batch_size\n            else:\n                n_iter = whole_size // batch_size + 1\n                \n            all_events = set()\n            for batch_i in range(n_iter):\n                batch = image[batch_i * batch_size:(batch_i + 1) * batch_size]\n                if batch.ndim == 3:\n                    batch = batch.unsqueeze(0)\n\n                batch = batch.to(device)\n                with torch.no_grad():\n                    prediction = model(batch)\n                    proba = torch.square(prediction.view(batch.size(0), -1).sigmoid()).cpu().numpy()\n                    prediction = model2(batch)\n                    proba += torch.square(prediction.view(batch.size(0), -1).sigmoid()).cpu().numpy()\n                    prediction = model3(batch)\n                    proba += torch.square(prediction.view(batch.size(0), -1).sigmoid()).cpu().numpy()\n                    \n                    proba = proba / 3\n                    proba = np.sqrt(proba)\n                    \n                events = proba >= threshold\n                for i in range(len(events)):\n                    event = events[i, :]\n                    labels = np.argwhere(event).reshape(-1).tolist()\n                    for label in labels:\n                        all_events.add(label)\n                        \n            labels = list(all_events)\n        if len(labels) == 0:\n            prediction_dict[row_id] = \"nocall\"\n        else:\n            labels_str_list = list(map(lambda x: INV_BIRD_CODE[x], labels))\n            label_string = \" \".join(labels_str_list)\n            prediction_dict[row_id] = label_string\n    return prediction_dict\n\n\nunique_audio_id = test_df.audio_id.unique()\nwarnings.filterwarnings(\"ignore\")\nprediction_dfs = []\nfor audio_id in progress_bar(unique_audio_id):\n    # Load the spectrogram\n    signal, sample_rate = torchaudio.load(os.path.join(test_dir, audio_id+'.mp3'))\n\n    signal = signal.mean(0).cuda()\n    signal = torchaudio.transforms.Resample(orig_freq=sample_rate, new_freq=args['std_sr'])(signal)\n    print('signal shape',signal.shape)\n    spectrogram = torch.log((torch.stack([f1(signal), f2(signal), f3(signal)], axis=0)) + 1e-8).half().float().cpu()\n    print('spectrogram after log and torchstack',spectrogram.shape)\n\n    test_df_for_audio_id = test_df.query(\n        f\"audio_id == '{audio_id}'\").reset_index(drop=True)\n    dataset = TestDataset(test_df_for_audio_id, spectrogram)\n    prediction_dict = prediction_for_clip(dataset=dataset, threshold=args['threshold'])\n    row_id = list(prediction_dict.keys())\n    birds = list(prediction_dict.values())\n    prediction_df = pd.DataFrame({\n        \"row_id\": row_id,\n        \"birds\": birds\n    })\n    prediction_dfs.append(prediction_df)\n\nprediction_df = pd.concat(prediction_dfs, axis=0, sort=False).reset_index(drop=True)\nprediction_df['birds'].value_counts()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"prediction_df.to_csv('submission.csv', index=False)\nprediction_df","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"markdown","source":"for i in range(len(prediction_df)):\n    if 'site_3' in prediction_df['row_id'].values[i]:\n        prediction_df.birds.values[i] = 'nocall'\nprediction_df.to_csv('submission.csv', index=False)\nprediction_df","execution_count":null}],"metadata":{"kernelspec":{"language":"python","display_name":"Python 3","name":"python3"},"language_info":{"pygments_lexer":"ipython3","nbconvert_exporter":"python","version":"3.6.4","file_extension":".py","codemirror_mode":{"name":"ipython","version":3},"name":"python","mimetype":"text/x-python"}},"nbformat":4,"nbformat_minor":4}