Learning Lightweight Lane Detection CNNs by Self Attention Distillation

Training deep models for lane detection is challenging due to the very subtle and sparse supervisory signals inherent in lane annotations. Without learning from much richer context, these models often fail in challenging scenarios, e.g., severe occlusion, ambiguous lanes, and poor lighting conditions... (read more)

PDF Abstract ICCV 2019 PDF ICCV 2019 Abstract

Results from the Paper


TASK DATASET MODEL METRIC NAME METRIC VALUE GLOBAL RANK RESULT BENCHMARK
Lane Detection BDD100K ENet-SAD Accuracy 36.56% # 1
Lane Detection CULane ENet-SAD F1 score 70.8 # 14
Lane Detection TuSimple ENet-SAD Accuracy 96.64% # 2
F1 score 95.92 # 13

Methods used in the Paper


METHOD TYPE
ReLU
Activation Functions
Average Pooling
Pooling Operations
Residual Connection
Skip Connections
Global Average Pooling
Pooling Operations
Bottleneck Residual Block
Skip Connection Blocks
Residual Block
Skip Connection Blocks
Kaiming Initialization
Initialization
ResNet
Convolutional Neural Networks
Dilated Convolution
Convolutions
1x1 Convolution
Convolutions
Batch Normalization
Normalization
Max Pooling
Pooling Operations
Convolution
Convolutions
ENet Dilated Bottleneck
Image Model Blocks
ENet Bottleneck
Image Model Blocks
ENet Initial Block
Image Model Blocks
SpatialDropout
Regularization
PReLU
Activation Functions
ENet
Semantic Segmentation Models