Tiny yolo

• Use pretrained weights
• Architecture: InputLayer, model_1, Conv2D(Detection), Reshape, InputLayer, Lambda
• Time: more than 11s (not accurate)
• Always get Allocation of 37748736 exceeds 10% of system memory
• Parameters: Total params: 15,770,510 Trainable params: 15,764,398 Non-trainable params: 6,112

MobileNet

• Get error, axes don't match array
• Figure out the error
• Train on google cloud (keras_tf) shuyangan_seu@deep-1:~/deep_learning/keras-yolo2$ python3 train.py -c config.json Using TensorFlow backend. Seen labels: {'raccoon': 217} Given labels: ['raccoon'] Overlap labels: {'raccoon'} 2018-06-16 23:19:34.106656: I tensorflow/core/platform/cpu_feature_guard.cc:140] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA 2018-06-16 23:19:34.248386: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:898] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero 2018-06-16 23:19:34.248794: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1356] Found device 0 with properties: name: Tesla K80 major: 3 minor: 7 memoryClockRate(GHz): 0.8235 pciBusID: 0000:00:04.0 totalMemory: 11.17GiB freeMemory: 11.09GiB 2018-06-16 23:19:34.248823: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1435] Adding visible gpu devices: 0 2018-06-16 23:19:34.545609: I tensorflow/core/common_runtime/gpu/gpu_device.cc:923] Device interconnect StreamExecutor with strength 1 edge matrix: 2018-06-16 23:19:34.545661: I tensorflow/core/common_runtime/gpu/gpu_device.cc:929] 0 2018-06-16 23:19:34.545670: I tensorflow/core/common_runtime/gpu/gpu_device.cc:942] 0: N 2018-06-16 23:19:34.545959: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1053] Created TensorFlow device (/job:localhost/replica:0/task:0/device:GPU:0 with 10751 MB memory) -> physical GPU (device: 0, name: Tesla K80, pci bus id: 0000:00:04.0, compute capability: 3.7) (13, 13)

---

Layer (type) Output Shape Param # Connected to

input_1 (InputLayer) (None, 416, 416, 3) 0

---

model_1 (Model) (None, 13, 13, 1024) 3228864 input_1[0][0]

---

DetectionLayer (Conv2D) (None, 13, 13, 30) 30750 model_1[1][0]

---

reshape_1 (Reshape) (None, 13, 13, 5, 6) 0 DetectionLayer[0][0]

---

input_2 (InputLayer) (None, 1, 1, 1, 10, 0

---

lambda_1 (Lambda) (None, 13, 13, 5, 6) 0 reshape_1[0][0]
input_2[0][0]

Total params: 3,259,614 Trainable params: 3,237,726 Non-trainable params: 21,888