Day-6 Computer Vision Learning 𝗗𝘆𝗻𝗮𝗺𝗶𝗰 𝗥-𝗖𝗡𝗡: Towards High-Quality Object Detection via Dynamic Training Published in #ECCV2020 𝗗𝘆𝗻𝗮𝗺𝗶𝗰 𝗥-𝗖𝗡𝗡: ------------------------------------------------------------------- 𝗔𝗺𝗮𝘇𝗶𝗻𝗴 𝗥𝗲𝘀𝗲𝗮𝗿𝗰𝗵: https://lnkd.in/eJH-4dA 𝗣𝘆𝘁𝗼𝗿𝗰𝗵 𝗩𝗲𝗿𝘀𝗶𝗼𝗻 : 1. https://lnkd.in/e9_KXU9 2. https://lnkd.in/eHpYDnN ------------------------------------------------------------------- Component Used in Paper: 🔹 Dynamic SmoothL1 Loss: Loss function in object detection where we change the shape of the loss function to gradually focus on high-quality samples 🔹 Non-Maximum Suppuration: Select the best bounding box for an object and reject or “suppress” all other bounding boxes. 🔹 RPN: A Region Proposal Network, or RPN, is a fully convolutional network that simultaneously predicts object bounds and objectness scores at each position. 🔹 Softmax: Softmax is an activation function that outputs the probability for each class and these probabilities will sum up to one. ------------------------------------------------------------------- #computervision #artificialintelligence #machinelearning #deeplearning #opencv #innovation #india
Oracle•105K followers
5yDynamic R-CNN to adjust the label assignment criteria (IoU threshold) and the shape of the regression loss function (parameters of SmoothL1 Loss) automatically based on the statistics of proposals during training. This dynamic design makes better use of the training samples and pushes the detector to fit more high-quality samples. Specifically, our method improves upon ResNet-50-FPN baseline with 1.9% AP and 5.5% AP90 on the MSCOCO dataset with no extra overhead.