| # 199 line in PostProcessIR_v11.py file | |
| # 요게 원본 소스코드 | |
| # 고정된 min, max 온도에 맞게 컬러를 scaling하지 않고 | |
| # 이미지별 min, max 온도 값을 기준으로 컬러를 scaling함. | |
| def raw_to_8bit(data): | |
| cv2.normalize(data, data, 0, 65535, cv2.NORM_MINMAX) | |
| np.right_shift(data, 8, data) | |
| return cv2.cvtColor(np.uint8(data), cv2.COLOR_GRAY2RGB) |
Minimal example: transcode from MP3 to WMA:
ffmpeg -i input.mp3 output.wma
You can get the list of supported formats with:
ffmpeg -formats
Convert WAV to MP3, mix down to mono (use 1 audio channel), set bit rate to 64 kbps and sample rate to 22050 Hz:
| net = tf.layers.conv2d(inputs, 64, 11, 4, padding="VALID", activation=tf.nn.relu, name="conv1") | |
| net = tf.layers.max_pooling2d(net, 3, 2, name="max_pool1") | |
| attention_mask = tf.layers.conv2d(net, 1, 1, padding="SAME") | |
| shape = attention_mask.get_shape() | |
| features = tf.reshape(tf.transpose(attention_mask, [0, 3, 1, 2]), | |
| [config.batch_size * int(shape[3]), int(shape[1]) * int( | |
| shape[2])]) | |
| spatial_softmax = tf.nn.softmax(features) |
| import numpy as np | |
| # input x | |
| x = np.array([[1], [2], [3], [4], [5], [6], [7], [8], [9], [10]]) | |
| print("input x: %f" % x.var()) | |
| # high variance w1 | |
| w1 = np.random.randn(10, 10) | |
| w1[0][0] = 100 | |
| print("parameter w1 high variance : %f" % w1.var()) |
| import numpy as np | |
| # high variance input x1 | |
| x1 = np.array([[1], [2], [3], [4], [5], [6], [7], [8], [9], [100]]) | |
| # low variance input x2 | |
| x2 = np.array([[1], [2], [3], [4], [5], [6], [7], [8], [9], [10]]) | |
| print("input x1 high variance : %f" % x1.var()) | |
| print("input x2 low variance : %f" % x2.var()) | |
| # gaussian distribution weights w |
| # import tensorflow as tf | |
| import numpy as np | |
| # x1 = np.random.random_integers(5, size=(3,2)) | |
| x1 = np.array([[1], [2], [3], [4], [5], [6], [7], [8], [9], [100]]) | |
| x2 = np.array([[1], [2], [3], [4], [5], [6], [7], [8], [9], [10]]) | |
| print(x1.var(), x2.var()) | |
| w1 = np.random.randn(10, 10) | |
| w2 = np.random.randn(10, 10) | |
| w2[0][0] = 100 |
| def plotNNFilter(units): | |
| filters = units.shape[3] | |
| plt.figure(1, figsize=(20,20)) | |
| n_columns = 6 | |
| n_rows = math.ceil(tmp.shape[3] / n_columns) | |
| for i in range(0, filters): | |
| plt.subplot(n_rows, n_columns, i+1) | |
| plt.title('Filter ' + str(i)) | |
| plt.imshow(units[0,:,:,i], interpolation="nearest", cmap="gray") |