arxiv: https://arxiv.org/pdf/1704.04861.pdf key points focus on optimizing for latency, small networks. use depthwise separable convolutions, to reduce computation as much as possible further reduce size of models based on width/resolution multiplier, but at the cost of accuracy depthwise separable convolution This is a combination of depthwise convonlution + pointwise convolution. Read more…
- model to predict depth map
- maximize speed by making it light as possible
- focus not only on encoder network but also on decoder network for speed improvement
- mobilenet for encoder, nearest-neighbor interpolation + NNConv5 for decoders, use skip connection, use depthwise separable convolution where ever possible, do network pruning, use TVM compiler stack to optimize depthwise separable convolution which is not optimized in populate DL frameworks.
arxiv link: https://arxiv.org/abs/1911.09070 key points multi scale with weighted bi-directional fpn. model scaling. compound scaling method, which jointly scales up resolution/depth/width for all backbone, feature network, box/class prediction network. use efficientnet backbone Bi-directional FPN Here are the key points of bi-directional FPN enhancement from PANet with some modifications remove nodes Read more…
while looking through some deep learning implementation code, I found some strange list indexing in python code. Here is an example. It was using None as an index when indexing a python object. First I thought is this indexing None on a python list and tested it. But it gave Read more…
algorithm to find if closed path is rotating in right or left direction(finding right hande rule direction of closed loop)
A very simple mathmatical method exist to solve this problem. https://mathworld.wolfram.com/PolygonArea.html here is an interactive example: https://demonstrations.wolfram.com/SignedAreaOfAPolygon by calculating the signed area of a polygon, one can easily determine if a 2D directed path is rotating in the right or left direction.