From 1415a798fe214432adc68c90ae323accf2071359 Mon Sep 17 00:00:00 2001 From: Glenn Jocher Date: Sat, 17 Nov 2018 12:54:44 +0100 Subject: [PATCH] updates --- models.py | 20 ++++++++++---------- utils/utils.py | 8 ++++---- 2 files changed, 14 insertions(+), 14 deletions(-) diff --git a/models.py b/models.py index 61c66dc1..b9e3f88d 100755 --- a/models.py +++ b/models.py @@ -123,16 +123,16 @@ class YOLOLayer(nn.Module): y = torch.sigmoid(p[..., 1]) # Center y # Width and height (yolo method) - # w = p[..., 2] # Width - # h = p[..., 3] # Height - # width = torch.exp(w.data) * self.anchor_w - # height = torch.exp(h.data) * self.anchor_h + w = p[..., 2] # Width + h = p[..., 3] # Height + width = torch.exp(w.data) * self.anchor_w + height = torch.exp(h.data) * self.anchor_h # Width and height (power method) - w = torch.sigmoid(p[..., 2]) # Width - h = torch.sigmoid(p[..., 3]) # Height - width = ((w.data * 2) ** 2) * self.anchor_w - height = ((h.data * 2) ** 2) * self.anchor_h + # w = torch.sigmoid(p[..., 2]) # Width + # h = torch.sigmoid(p[..., 3]) # Height + # width = ((w.data * 2) ** 2) * self.anchor_w + # height = ((h.data * 2) ** 2) * self.anchor_h # Add offset and scale with anchors (in grid space, i.e. 0-13) pred_boxes = FT(bs, self.nA, nG, nG, 4) @@ -174,8 +174,8 @@ class YOLOLayer(nn.Module): # lconf = k * BCEWithLogitsLoss(pred_conf[mask], mask[mask].float()) lconf = (k * 10) * BCEWithLogitsLoss(pred_conf, mask.float()) - lcls = (k / 10) * CrossEntropyLoss(pred_cls[mask], torch.argmax(tcls, 1)) - # lcls = k * BCEWithLogitsLoss(pred_cls[mask], tcls.float()) + # lcls = (k / 10) * CrossEntropyLoss(pred_cls[mask], torch.argmax(tcls, 1)) + lcls = (k * 10) * BCEWithLogitsLoss(pred_cls[mask], tcls.float()) else: lx, ly, lw, lh, lcls, lconf = FT([0]), FT([0]), FT([0]), FT([0]), FT([0]), FT([0]) diff --git a/utils/utils.py b/utils/utils.py index 55e7e64a..f62eca34 100755 --- a/utils/utils.py +++ b/utils/utils.py @@ -259,12 +259,12 @@ def build_targets(pred_boxes, pred_conf, pred_cls, target, anchor_wh, nA, nC, nG ty[b, a, gj, gi] = gy - gj.float() # Width and height (yolo method) - # tw[b, a, gj, gi] = torch.log(gw / anchor_wh[a, 0]) - # th[b, a, gj, gi] = torch.log(gh / anchor_wh[a, 1]) + tw[b, a, gj, gi] = torch.log(gw / anchor_wh[a, 0]) + th[b, a, gj, gi] = torch.log(gh / anchor_wh[a, 1]) # Width and height (power method) - tw[b, a, gj, gi] = torch.sqrt(gw / anchor_wh[a, 0]) / 2 - th[b, a, gj, gi] = torch.sqrt(gh / anchor_wh[a, 1]) / 2 + # tw[b, a, gj, gi] = torch.sqrt(gw / anchor_wh[a, 0]) / 2 + # th[b, a, gj, gi] = torch.sqrt(gh / anchor_wh[a, 1]) / 2 # One-hot encoding of label tcls[b, a, gj, gi, tc] = 1